Chapter Nine

Is There Immediate Justification?

You have immediate justification for believing that p if, and only if, your justification for believing that p does not depend on any justification you have for believing some other proposition, q. James Pryor defends the claim that such justification exists. According to him, the best argument in its support is not the regress argument, according to which unacceptable alternatives – justification coming from an infinite regress of reasons, from a circular chain of reasons, or from initial beliefs that are not themselves justified – lend support to the conclusion that there must be immediate justification. Rather, Pryor argues, the best way to defend the existence of immediately justified beliefs is to cite examples of such beliefs, such as “I am tired” or “I have a headache.” Pryor provides additional support for immediate justification by criticizing what he calls the “Master Argument for Coherentism.” In his essay, Juan Comesaña raises a serious problem for the view that there are immediately justified beliefs. He argues that a view like Pryor’s is committed to four principles each of which is, by itself, exceedingly plausible. Unfortunately, the four principles are inconsistent. In response to Comesaña, Pryor singles out and rejects what Comesaña calls the Entailment Principle. Comesaña, in turn, defends the principle in his reply to Pryor.

There Is Immediate Justification James Pryor

1 Justification

I want to talk about a certain epistemic quality that I call “justification,” and inquire whether that quality can ever be had “immediately” or “non-inferentially.” Before we

Contemporary Debates in Epistemology, Second Edition. Edited by Matthias Steup, John Turri, and Ernest Sosa. © 2014 John Wiley & Sons, Inc. Published 2014 by John Wiley & Sons, Inc. get into substantive issues, we need first to agree about what epistemic quality it is we’ll be talking about, and then we need to clarify what it is to have that quality immediately or non-inferentially. When I say I call this epistemic quality “justification,” you are liable to think, “Oh I know what that is.” You may. But experience has taught me that different philosophers use and understand the term “justification” differently, even before they start spinning substantive theories about what “justification” amounts to. So we should proceed cautiously. You may use the term “justification” to describe the same epistemic quality as I do; or you may use it to describe some different status or quality. You may use some other term, or no term at all, to describe the quality I call “justification.” I say that you have justification to believe P iff you are in a position where it would be epistemically appropriate for you to believe P, a position where P is epistemically likely for you to be true. I intend this to be a very inclusive epistemic status.1 Some philosophers say you can know P without “having any justification” for your belief. We can assume that whenever a subject knows P, she will be in a position where it would be epistemically appropriate to believe P. So on my usage, whoever knows P has justification to believe P. (Perhaps she has that justification because she knows.) The philosophers who say otherwise are using “having justification” to mean something different, or more specific, than the epistemic status I am using it to mean. The same goes for philosophers who say a belief can be epistemically appropriate, and so play a role in justifying other beliefs, though you do not “have any justification” for it. On my usage, all it means to have justification to believe something is that it is appropriate for you to believe it, and to rely on that belief in forming other beliefs. Some philosophers call this epistemic status “entitlement” or “warrant,” rather than “justification.” For the sake of a shared discussion, though, we need to fix on a single terminology. If there is some state or condition you are in in virtue of which you have justification to believe P, I’ll call it a “justification-making condition,” or a justification maker for short. This is a condition that makes it epistemically appropriate (or more appropriate) for you to believe P, rather than disbelieve P or suspend judgment. It is a truth maker for your having justification to believe P. (Firth, 1964, called these conditions “warrant-increasing properties.”) We can say that conditions of this sort justify you in believing P. They are justifiers. (We will encounter a different way to understand talk of “justifiers” in section 6 below.) In what follows, it will be useful for us to distinguish between having justification to believe P, and actually appropriately holding a belief in P. To have justification to believe P, it is not important whether you actually do believe P (nor, if you do, why you do); there just have to be things that make believing P an appropriate attitude for you to have. To appropriately believe P more is required. You need to believe P; you need to have justification to believe P; and you also need to believe P on the right grounds. You need to believe it for reasons that make you have justification to believe it; you can’t believe it for other, bad reasons, or on a whim.2 There are further conditions as well: for instance, you need to be taking proper account of any evidence you have that tells against or undercuts your grounds for believing P. I describe another further condition in Pryor (2004). Only when all such conditions are met will your belief in P be appropriately held.

There Is Immediate Justification 203 2 Immediate Justification

Now that we have a grip on the notion of “justification,” let’s clarify what it means to talk about “immediate justification.” For some propositions, you have justification to believe them because other propositions you have justification to believe epistemically support them. For instance, suppose you look at the gas gauge of your car, and it appears to read “E.” So you have justification to believe:

(Gauge) The gas gauge reads “E.”

That, together with other things you justifiedly believe about your car, gives you justification to believe:

(Gas) Your car is out of gas.

(It is not important for our purposes whether you actually do believe (Gauge) or (Gas). Given your evidence, you ought to believe them.) In this example, your justification to believe (Gas) comes in part from the fact that you have justification to believe (Gauge). That is, having justification to believe the latter is part of what makes you have justification to believe the former. The justification you have in this example to believe (Gauge) does not in the same way come from your having justification to believe (Gas). (One mark of this is that evidence that undercut your justification to believe (Gauge) would ipso facto undercut your justification to believe (Gas); but not vice versa.) When your justification to believe P comes in part from your having justification to believe other, supporting propositions, I will say that those latter propositions mediate your justification to believe P. (This kind of justification is sometimes called “inferential” justification. We will encounter a second way in which justification can be “inferential” later.) When your justification to believe P does not come from your justification to believe other propositions, I will call it immediate. Some clarifications.3 First, the question whether your justification to believe P is mediate or immediate is a question about what kind of epistemic support you have for P. It is not a question about how much support you have: nothing in our definition requires immediately justified beliefs to be infallible or indefeasible. Nor is it a question about what psychological processes you have undergone. The support you have to believe P can be mediate (or “inferential”) even if you didn’t arrive at P by deriving or inferring it from other beliefs. Second, in order for you to have immediate justification to believe P, it is not required that your justification comes from nowhere, that there is nothing that makes you so justified. It is only required that what makes you justified doesn’t include having justification for other beliefs.4 There are various proposals about what can make one have immediate justification. For example, perhaps you are immediately justified in believing you feel tired because you do feel tired. Perhaps you are immediately justified in believing that tiredness is a mental state because you understand what tiredness is. And so on. It may be

204 James Pryor that there is no single correct account. Different propositions may be justified by different kinds of things. Third, the fact that you have immediate justification to believe P does not entail that no other beliefs are required for you to be able to form or entertain the belief that P. Having the concepts involved in the belief that P may require believing certain other propositions; it does not follow that any justification you have to believe P must be mediated by those other propositions.5 Fourth, justification is usually defeasible. What a justification maker for P gives you is provisional or prima facie justification to believe P; and that is what I am saying can be mediate or immediate. Whether it is all things considered appropriate for you to believe P will depend on what other evidence you possess, and whether it defeats the prima facie justification you have to believe P. Fifth, beliefs can be epistemically overdetermined. You can have immediate justification and independent mediate justification to believe the same thing. In some cases, your belief will be grounded on the facts that make you have immediate justification; in other cases it might be grounded on the facts that make you have mediate justification, or on both sets of facts. This shows that we should try to explain the notion of “grounding” in a way that permits beliefs to be both grounded and immediately justified – if that is possible. (In section 7 we will consider an argument that it is not possible.)

3 Why Believe in Immediate Justification?

Now that we have achieved some clarity about what immediate justification is, let’s ask why we should believe in it. The most famous argument for immediate justification is called the Regress Argument. Really this is not one argument but several; because philosophers do not always have the same regress in mind. Sometimes they have in mind a dialectical regress: to defend your belief that P in argument, you need to appeal to other beliefs, but then how do you defend those other beliefs? Sometimes they have in mind a grounding regress: your belief in P is grounded in such-and-such other beliefs, but then what grounds those other beliefs? Sometimes they have in mind a justification- making regress: what makes you have justification to believe P is, in part, your having justification to believe such-and-such other propositions, but then what makes you justified in believing those other propositions? Let’s focus on this third, justification-making regress. There are four possible ways for the regress to play out:

(i) The regress never ends. The justificatory chain goes on forever. (ii) What makes you justified in believing P is your having justification to believe other things, and … what makes you justified in believing some of them is your having justification to believe P. That is, the justificatory chain includes some closed loops. (iii) Eventually we get to a proposition you believe inappropriately, without having any justification for it. Though this belief is not itself justified, it is somehow able to justify you in believing further propositions.

There Is Immediate Justification 205 (iv) Eventually we get to a proposition you have justification to believe, but that justification does not come from your believing, or having justification to believe, any further propositions.6

The foundationalist argues that options (i) and (ii) are untenable; so we have to accept (iii) or (iv). On either of those options, a subject can have justification to believe some propositions, that does not come from her having justification to believe any other propositions.7 Though this Regress Argument is the most famous argument for immediate justification, I do not think it is the best argument. It has the same weakness as any argument by elimination: everything turns on whether the rejected options really are untenable. That is not a matter that can be quickly decided. In addition, the Regress Argument assumes that justificatory relations always have a linear, asymmetric nature; and some epistemologists deny that that is so. So I do not think the best argument for immediate justification is this Regress Argument. I think the best argument comes from considering examples. Suppose I feel tired, or have a headache. I am justified in believing I feel those ways. And there do not seem to be any other propositions that mediate my justification for believing it. What would the other propositions be? Suppose I raise my arm. I am justified in believing that I’m doing this in order to scare a fly. That is my reason for trying to raise my arm. Sometimes my reasons for acting are opaque to me and have to be carefully reconstructed. But not always. In cases like this one, my reasons can be immediately evident to me. There doesn’t seem to be anything else I am justified in believing, that makes me justified in believing my reason for trying to raise my arm is to scare a fly. What would the other beliefs be? I am imagining my grandmother. The way I am imagining her is sitting in her kitchen. Or at least, I believe it is. And it seems I could be justified in that belief. Again, it is hard to see what other propositions might mediate this justification. I think about a domino and a chessboard. It is obvious to me that the only way to wholly cover two spaces on the board is to place the domino horizontally or vertically. That is something I could have derived from geometric premises. But in this case I didn’t. I just immediately saw that it was true. In this case, too, my justification does not seem to be mediated by any further propositions.8 These and many other examples provide us with good candidates to be immediate justification. What we need to do is see whether such examples stand up to critical reflection.

4 The Master Argument for Coherentism

The main argument against immediate justification was historically directed at the Given Theory. That was a theory that offered a specific account of what some immediate justification makers looked like. The precise details of the Given Theory aren’t important for our inquiry.9 What is important is that the Given Theory is just one possible account among many of what gives us immediate justification. Fans of immediate justification are also free to give different accounts. The following map may be helpful:

206 James Pryor I make foundationalism a proper subclass of the view that there is immediate justification, because foundationalists also hold additional theses about the structure of your justification. One does not need to accept those additional theses, merely to believe that some beliefs are immediately justified. Coherentists deny that it is possible for any beliefs to be immediately justified. They say that justification always comes, at least in part, from your justification for other beliefs. I distinguish between pure and impure versions of coherentism. Pure coherentists claim that a belief can only be justified by its relations to other beliefs. Impure coherentists are willing to give some non-beliefs, such as perceptual experiences, a justifying role. They will just deny that those states are able to justify a belief all by themselves. They can only do so in cooperation with other justified beliefs. For instance, an impure coherentist might say that when it looks to you as if you have hands, and you have justification to believe that your visual experiences are reliable, those facts together can make you justified in believing that you have hands.10 Now, as I said, the main argument against immediate justification was historically directed at the Given Theory. This argument alleged that in order to be a justifier, you need to have certain characteristics, and that having those characteristics makes you be the sort of thing that itself needs justification. Here is a sample presentation, from BonJour:

The basic idea of givenness … is to distinguish two aspects of ordinary cognitive states, their capacity to justify other cognitive states and their own need for justification, and then to try to find a kind of state which possesses only the former aspect and not the latter – a state of immediate apprehension or intuition. But we can now see plainly that any such attempt is fundamentally misguided and intrinsically hopeless. For it is clear on reflection that it is one and the same feature of a cognitive state, namely, its assertive or at least representational content, which both enables it to confer justification on other states and also creates the need for it to be itself justified – thus making it impossible in principle to separate these two aspects. (BonJour, 1985, p. 78)

The characteristics BonJour cites are having “assertive” or representational content. (Sometimes it is claimed, in addition, that a justifier has to have conceptual content. We will return to that idea later.) In order to have these characteristics, the coherentist argues, a state would itself have to be a belief – or at least be sufficiently like a belief that it in turn needs justifying. So we aren’t going to find any immediate justifiers: conditions that are able to justify though they don’t themselves need justifying. That is the core of the familiar argument against the Given Theory. Notice that this argument really doesn’t have anything specifically to do with the Given Theory. If it works, it should work against any account of immediate justification. In fact, if it works, then the only things which can play any justifying role will be other beliefs (or belief-like states). So the argument threatens impure coherence theories no less than it does views that countenance immediate justification. It is not really an argument against the Given Theory, then. It is more an argument for pure coherentism. I think of this argument as the Master Argument for coherentism.11 One step in this Master Argument insists that only states with propositional content can be justifiers. Let’s call this:

There Is Immediate Justification 207 The Content Requirement In order to be a justifier, you need to have propositional content.

Why should we accept that requirement? Well, if a state doesn’t have propositional content, then it can’t stand in logical relations to beliefs. Davidson once complained:

The relation between a sensation and a belief cannot be logical, since sensations are not beliefs or other propositional attitudes. What then is the relation? The answer is, I think, obvious: the relation is causal. Sensations cause some beliefs and in this sense are the basis or ground of those beliefs. But a causal explanation of a belief does not show how or why the belief is justified. (Davidson, 1986, p. 311)

Of course Davidson is right that merely learning that some sensation S caused belief B does not show that B is justified, or that S is what makes it justified. However, it would be compatible with that that S does make B justified. Davidson wants to rule that possibility out. He thinks that sensation S cannot be what justifies B. So he must think that it is only by standing in logical relations to a belief that a state can justify the belief. He may be thinking: if a state doesn’t stand in logical relations to a belief, then why should it justify that belief as opposed to others? For example, let’s assume that headaches don’t have propositional content. (Some philosophers argue that all mental states have propositional content. For the sake of discussion, we will assume with Davidson that they are wrong.) Why then should a headache justify me in believing I have a headache, as opposed to I don’t have a headache or I am a headache or There are no such things as headaches? My headache itself wouldn’t logically support any of those propositions; so it is not clear why it should justify some of them but not the others. There is then some initial plausibility to the idea that in order to play the role of a justifier, a state has to be able to stand in logical relations; which it can only do if it has propositional content. (Objection: What about a lack of defeating evidence? That is not naturally thought of as a state with propositional content; but it does seem relevant to your justification. Reply: The Content Requirement should be understood as stating a necessary condition to be a prima facie justifier. The role defeating evidence plays is in settling a further question: When does prima facie justification get to become all things considered justification?) The other step in the Master Argument insists that if a state has propositional content, then it will be a belief or epistemically like a belief:

Only Beliefs Only beliefs (or other states that are epistemically like beliefs, and also require epistemic justification) have propositional content.

But wait a minute! Desires have propositional content, and they are not the sort of thing which could be, nor do they need to be justified – at least, not epistemically justified.12 The coherentist will respond: “That’s true, desires have propositional content and don’t need any epistemic justification. So the Only Beliefs premise as it stands is false. But desires aren’t capable of justifying beliefs, either! So they’re not a counterexample to the conclusion we want: that only beliefs can be justifiers. They just force us to be more specific about what features it is that enable a state to be a justifier. It takes more

208 James Pryor than just having propositional content. There are some ways of representing the proposition that P that purport to be saying how the world is, and other ways that don’t. When a state represents that P in the first way, we can say that the state assertively represents that P. Desires may represent that P in some sense, but they do not represent it assertively. Neither do states like imagining and entertaining. When you desire that P, or imagine that P, or entertain the thought that P, your mental state does not purport to be saying that P is (already) true. What our Master Argument should say is that, to be a justifier, a state needs to have propositional content, and it needs to represent that proposition assertively. States that don’t do that aren’t even purporting to say how the world is, so how could they play the role of justifiers? And in order to represent a proposition assertively, a state will have to be a belief, or else sufficiently like a belief that it needs justifying too.” Let’s revise the Master Argument as the coherentist proposes:

The Content Requirement (Revised) In order to be a justifier, you need to have propositional content, and you need to represent that proposition assertively.

Only Beliefs (Revised) Only beliefs (or other states that are epistemically like beliefs) represent propositions assertively. ∴ Only beliefs (or other states that are epistemically like beliefs) can be justifiers.

I want to mention two quick worries about this argument, and then dwell at length upon a third. First, even if this argument were sound, it is not clear that it would establish coherentism. The argument says that the only things that could justify are states like belief, that “require” justification. A foundationalist might agree that beliefs in some sense always “require” justification, but argue that they are still sometimes able to justify other beliefs even when they are not themselves justified. (This was option (iii) in the Regress Argument.) The other beliefs that got justified in this way would count as “immediately justified,” as I have defined it. Second, we might worry whether the coherentist is himself in a position to accept the Master Argument. After all, doesn’t the coherentist want facts about coherence to play a justifying role? Yet coherence is not itself a belief or a belief-like state. Here I think the coherentist can reply, “Notice that coherence is a property of the contents of your beliefs. Any set of beliefs having the same contents would be just as coherent. So it is OK to say that it is always your beliefs that are doing the justifying. It is just that certain sets of beliefs (those whose contents cohere well) justify more than others. Talk about the justifying role of coherence is shorthand for talk about which sets of beliefs justify and which don’t.” This seems to me a plausible line for the coherentist to take. There is a third worry that the coherentist cannot so easily finesse, however. This worry once again concerns the Only Beliefs premise. The problem is that many philosophers of mind these days think of experiences as having propositional content. To say that experiences have propositional content is not to say that experiences are beliefs. It can look to you as if P without your believing that P. Experiences and beliefs just have it in common that they both represent propositions.13 And both seem to represent

There Is Immediate Justification 209 propositions assertively; when they represent that P, they do so in a way that purports to say how the world is. So the Master Argument as we have it gives us no reason to exclude experiences from the ranks of justifiers.14 Yet, unlike beliefs, experiences aren’t the sort of thing which could be, nor do they need to be justified. Sure, beliefs about what experiences you have may need to be justified. But the experiences themselves do not. (If someone comes up to you and demands, “How dare you have that experience? What gives you the right?” what should you say?) So we see that, contrary to the Only Beliefs premise, states that assertively represent a proposition won’t always themselves require justification. Where does this leave things between foundationalists (or fans of immediate justification more generally) and proponents of the Master Argument? A foundationalist can just stop here. He can say, “Well, we’ve seen that experiences are a counterexample to the Only Beliefs premise; so even if we accept the Content Requirement, that poses no obstacle to letting experiences play the role of justifiers. They might play that role in cooperation with other beliefs, as the impure coherentist allows, or they might do it all by themselves. We have seen no good reason yet to think they can’t” (see, for example, Martin, 1993; Steup, 2001). Some foundationalists do stop there. They are happy to accept the Content Requirement. But that is a pretty demanding constraint to put on what can be a justifier. And in fact, if you think about it, the Content Requirement will be well motivated only if an even more demanding constraint is. Let me explain. The coherentist denies that the mere presence of a headache or a desire can justify you in believing anything, because these states don’t assertively represent anything. They can’t even justify you in believing that you have those states. But then neither should the mere presence of a belief or an experience justify you in believing you have them, either. For although beliefs and experiences do have propositional contents that they assertively represent, those contents are playing no role in the justificatory relation here envisaged. It seems just an unfair prejudice to allow states that assertively represent propositions to justify the belief that you are in those states, but deny the same ability to other states. We should either allow this justifying relation in every case, or prohibit it in every case. The coherentist will try to prohibit it. They will say, “Look it’s not just that the justifying state needs to have a content. The content needs to be in some sense what does the justifying.” What that means, I guess, is that the content of the justifying state needs to imply, or inductively or abductively support, or stand in some other suitable evidential relation to the content it justifies. Let’s give this idea a name:

Premise Principle The only things that can justify a belief that P are other states that assertively represent propositions, and those propositions have to be ones that could be used as premises in an argument for P. They have to stand in some kind of inferential relation to P: they have to imply it or inductively support it or something like that.

The contents of your beliefs and experiences will ordinarily not imply that you are in those states; so this principle will prevent your beliefs and experiences from directly justifying the belief that you are in them.15 I think this Premise Principle is the real intuitive force behind the coherentist’s Content Requirement. Recall the Davidson quote from earlier. He was saying that to be

210 James Pryor a justifier, you have to stand in logical relations to the beliefs you justify. I think this is what he really had in mind. Here are some other authors also giving voice to the Premise Principle. We cannot really understand the relations in virtue of which a judgment is war- ranted except as relations within the space of concepts: relations such as implication or probabilification, which hold between potential exercises of conceptual capacities (McDowell, 1994, p. 7).

A reason for S’s believing that P is a fact about that person which makes her believing that thing intelligible from the point of view of rationality. If this is to happen then the selected fact about S must be somehow related to (her) believing that P. And since this relation is to make her believing that P intelligible from the point of view of rationality, it is necessarily a relation which obtains in virtue of the correctness of some kind of reasoning. That is to say, successfully giving such a reason makes essential reference to the premise of an inference of some kind, whose conclusion is appropriately related, most likely by identity, to the content of the belief for which the reason is being given. (Brewer, 1999, p. 154; see also pp. 150–151)

The Premise Principle says that all justifying relations between states hold in virtue of “inferential relations” between their contents. We have to be sure we understand this properly. First, the Premise Principle states a constraint on what can give you justification to believe things; it does not concern your actual beliefs or reasoning processes. It can allow that we often form beliefs that are supported by “inferential relations” without engaging in any inference. It can also respect the difference that Harman (1986) emphasized between logic and reasoning. Harman said that just because you believe some premises that together imply P, it doesn’t follow that it would be reasonable for you to infer P. It might, for example, be more reasonable to give up your belief in some of the premises. The Premise Principle can allow this. It is only trying to explicate the nature of prima facie justificatory relations. According to the Premise Principle, those always consist in “inferential relations” between contents. If your beliefs in some premises together imply P, and you have prima facie justification for those beliefs, then you have prima facie justification to believe P. But it might be unreasonable for you to infer P on those grounds, for example, if you also have other evidence that tells against P, or that undercuts some of your prima facie justification for the premises. Although there is a sense in which the Premise Principle claims all prima facie justificatory relations are “inferential,” this is not the same as saying that all justification is “inferential” in the sense of being mediated. The coherentists would like to argue from the Premise Principle to that conclusion. But the Premise Principle by itself doesn’t say it. Without supplementation, the Premise Principle would allow experiences to justify beliefs. For instance, an experience as of your having hands could justify the belief that you have hands. And justification of this sort would count as immediate. It is just that, according to the Premise Principle, the experience is able to justify that belief because of the “inferential relations” its content stands in to the content of the belief. (In this case, the “inferential relation” is straightforward: the experience’s content is the same as the content of the belief.) So the Premise Principle

There Is Immediate Justification 211 doesn’t imply what people usually have in mind when they say that our perceptual justification is “inferential” or mediated. That is a view according to which we are in the first place justified in believing we have certain experiences, and then that justification for beliefs about our experiences is an essential part of what justifies our beliefs about the external world. The Premise Principle doesn’t imply that; it would allow our merely having experiences with the right sorts of contents to justify beliefs about the external world. Nor is the Premise Principle implied by the view that perceptual justification is mediated. For the latter view says nothing about what justifies our beliefs about our experiences. Perhaps, contra the Premise Principle, we are justified in those beliefs merely by virtue of having the experiences. So there is no straightforward relation between the Premise Principle and the question whether perceptual justification is mediated or immediate. The relation between the Premise Principle and foundationalism more generally is also complicated. As I said before, some Foundationalists are happy to accept the coherentist’s Content Requirement, and the Premise Principle that motivates it. They just argue that there can be justifiers that satisfy the Premise Principle but aren’t beliefs. Those arguments interact in interesting ways with the question whether experiences have “conceptual content.” Some philosophers combine the Premise Principle with the view that you can only have the required type of “inferential relation” when both relata have conceptual content. That seems to be the view of Sellars (1963), McDowell (1994, lectures 1 and 3; 1998),16 and Brewer (1999, chapter 5). The coherentists will go on to argue that experiences don’t have conceptual content and so cannot be justifiers. McDowell and Brewer, on the other hand, think that experiences are justifiers; so they argue that experiences do have conceptual content after all. Others have argued that experiences can stand in the “inferential relations” the Premise Principle requires even if their content is not conceptual (see e.g., Heck, 2000; Peacocke, 2001).17 Personally, I am not really sure what “conceptual content” is; so I won’t enter into this debate. I just wanted to call attention to the role the Premise Principle plays in it. We have been looking at the Master Argument for coherentism, and considering whether it succeeds in ruling out the possibility of immediate justification. I said that even if the Master Argument were sound, it might still be possible for unjustified beliefs to do some justifying; and that would be a kind of immediate justification. We have also seen that one can accept the Premise Principle and still say that experiences justify; that will be another kind of immediate justification. So one does not need to reject the Premise Principle, to believe in immediate justification. Nonetheless, many foundationalists will want to reject the Premise Principle. It doesn’t exclude the very possibility of immediate justification; but it does impose quite a demanding constraint on what can be an immediate justifier. Many foundationalists believe in justification-making facts that violate that constraint. For instance, many foundationalists want to allow facts about what sensations you are having, or facts about what mental activities you are engaging in, to count as justifiers. Some want to allow facts about what is required to possess certain concepts to play a justifying role. Some say facts about how reliable you are, or facts about whether your cognitive faculties are functioning properly, or facts about what beliefs are irresistible, can play a justifying role. And so on. Each of these facts concerns matters that go beyond what

212 James Pryor assertive states you are in; so the Premise Principle would exclude them all from being justifiers. I think, then, that it would be valuable to consider whether the Premise Principle is really well motivated. The rest of this essay will consider some arguments on its behalf. There is one type of argument that I won’t consider. Those are “arguments from the trenches”: arguments of the form “Theory So-and-So gives the correct substantive account of justification; and that theory only postulates justifiers of the sort the Premise Principle permits.” Assessing any argument of that type would require examining the pros and cons of different theories of justification, and determining whether Theory So-and-So really is an adequate theory. That is well beyond the scope of this essay. Instead, I will look at arguments that try to establish the Premise Principle “from on high,” before we have decided upon a substantive theory of justification. I am going to argue that no argument of that sort succeeds. Hence, I think we have no reason to give the Premise Principle any authority when we are choosing among competing theories of justification. If I am right, that will clear the way for the many foundationalist theories that postulate immediate justifiers that violate the Premise Principle.18

5 Avoiding Arbitrariness

Our first argument for the Premise Principle is one that I have already mentioned. Consider a state without propositional content, like your headache. Since it has no propositional content, this state can’t stand in logical relations to any beliefs. So why should it justify any one belief as opposed to others? Why should it justify the belief I have a headache, as opposed to I don’t have a headache, or any other belief? What can the foundationalist say to make the justifying relations he postulates non-arbitrary? Chisholm provides a nice example of what is being objected to here. His view was that having an experience would give one prima facie justification for (or “tend to make evident”) certain beliefs (see e.g., Chisholm, 1989). The justifying relations Chisholm postulated struck many philosophers as ad hoc. That impression was particularly forceful because of the model of experience Chisholm worked with. When we describe an experience as one of sensing or being-appeared-to squarely, that is supposed to be a description of the experience’s intrinsic phenomenal quality. We might naturally take such an experience to be one in which we are perceiving an external square; but “being-appeared-to squarely” isn’t a description of what external objects we seem to be perceiving. It doesn’t mean “having an experience that represents that there is a square.” Chisholm didn’t think experiences had representational content. Hence, the justifying relations he postulated seemed to lack any principled motivation. Why should sensing squarely justify beliefs about squares, rather than beliefs about squirrels? Many think Chisholm had no adequate answer to this question; so his position seems arbitrary and unsatisfying. I certainly agree that epistemologists should give principled, non-arbitrary rationales for the justifying relations they postulate. However, I see no reason to think that they will have to appeal to propositional contents to do it. A foundationalist might attribute other kinds of structure to some of his justifiers. On certain theories of events, for example, events have something like a logical structure. The event of my

There Is Immediate Justification 213 having a headache has a logical structure akin to the structure of the proposition that I have a headache. Wouldn’t these structures be enough to enable the foundationalist to avoid the charge of arbitrariness?19 This is just one option for a foundationalist to pursue. It is hard to draw any general assessment, until we see how the details work out. But the idea that only the Premise Principle can save us from arbitrariness seems unwarranted to me.

6 Evidence and Reasons

Where I have been talking about “justification makers” or “justifiers,” some other philosophers will talk about “evidence” or “reasons”; and there are several arguments that the latter notions have to conform to the Premise Principle. This might be thought to show that justification makers also have to conform to the Premise Principle. Recall that a “justification maker” is defined to be whatever makes it epistemic appropriate for you to believe some propositions rather than others. We should be open to the possibility that terms like “evidence” and “reasons” do not express exactly that notion. They may express different, or more specific, notions. For instance, some philosophers argue that “reasons” and “evidence” have to be the sort of thing that can probabilify a hypothesis, and hence, that the hypothesis can have a probability conditional on. Hypotheses also have to be able to explain our evidence, be inconsistent with our evidence, and so on. All of these roles require evidence to have propositional content. So how could states without propositional content justify or be evidence? (See Williamson, 2000, pp. 194–197; Plantinga, 2001, p. 62.) In response, I say: let’s not assume too quickly that “evidence” and “justifier” are perfect synonyms. Consider that we sometimes use the terms “belief” and “desire” to refer to propositions that one believes and desires, rather than to one’s states of believing or desiring them. Similarly, I think, sometimes we use “evidence” to refer to propositions that are evident to one, rather than to the states that make them evident. In other words, we use “evidence” to refer not to our justification makers, but rather to the propositions that they (most directly) justify. We call those propositions “our evidence” because they can serve as evidence for further reasoning. This diagnosis would permit things like headaches, that do not themselves have propositional content, to be justification makers – so long as what they give one justification to believe is a proposition. A second argument says that we ordinarily understand “justifications” for a belief to be arguments that support the belief. If you have reasons for your belief, they should be considerations you could in principle cite, or give, to someone who doubted or chal- lenged the belief. You can’t give someone else a non-propositional state like a headache (at least, not in the relevant sense); you can only give them premises and arguments that inferentially support your belief. This seems to show that justifications and reasons are limited to things permitted by the Premise Principle. (See, for example, McDowell, 1994, pp. 165–166.) There may be a notion of “a reason” that these remarks properly articulate. We can call it the dialectical notion of a reason. I want to emphasize, though, that that notion is different from the notion of a justification maker that I have been employing in this paper, and that the Premise Principle is meant to be formulated in terms of.

214 James Pryor It is useful here to distinguish two construals of the verb “justify.” On the first con- strual, “justifying” a belief in P is a matter of proving or showing the belief to be just (or reasonable or credible). This is something that a person does, by giving some argument in support of that belief. (Here we can include both arguments whose conclusion is P, and arguments whose conclusion is that your belief in P is epistemically appropriate, or is likely to be true.) By extension, we can also talk about things justifying beliefs; in this extended sense, a thing counts as justifying a belief if it is something you are in a position to use to prove or show your belief to be just. Such things would be “reasons” in the dialectical sense articulated above. To be explicit, let’s call these things justification showers. There is also a second way to construe the verb “justify,” which sees it as akin to the verbs “beautify” and “electrify.” When a combination of light and color beautifies a room, it is not proving that the room is beautiful; rather, it is making the room beautiful. Similarly, on this understanding, justifying a belief is a matter of making a belief just or reasonable, rather than a matter of showing the belief to be just. That is how I understand the notion of a justification maker.20 No doubt there are some interesting connections between justification making and justification showing. But they are two different notions; so we should not assume that their extensions will coincide. It needs argument to show that nothing is eligible to be a justification maker unless it can also be a justification shower. Until we have such an argument, the fact that justification showers always conform to the Premise Principle should not persuade us that justification makers must do so as well. Even if we manage to separate the notion of a justification maker from the dialectical notion of “a reason,” I expect proponents of the Premise Principle will still insist there is enough of a connection between justification and “reasons” for constraints on the latter to support some restrictions on the former. For instance, they can observe that there is a difference between reasons there are to believe P – where these include reasons not now available to you – and reasons you have to believe P. For example, one reason there is to believe you will soon be sick is the fact that you just drank poison. But if you are unaware of that fact, then it is not a reason you have. For something to be a reason you have, for it to justify you in believing P, it has to be in some sense epistemically available to you. It has to be the sort of thing you could take as a reason. When it is not available to you – for example, when you are not in a state that assertively represents it, and so not in a position to appeal to it in arguing for P – then it may be a reason to believe P, but it won’t be a reason you have. For anyone with “internalist” sympathies, these reflections should apply to justifiers just as much as they do to reasons.21 I think it is right to distinguish between things such that you would be justified in believing P, if you were aware of them, and things that do justify you in believing P. I think it is also right that if something justifies you, then it has to be in some sense “available” to you. But I think it would be wrong to assume that this kind of “availability” requires you to be in representational states. As I understand the notion of “availability,” it is correlative to the notion of a ground. A justifier is available to you at a given time – it will be something you can “take as” a reason – if it is something that could then ground a belief of yours. If the foundationalist can make sense of beliefs being grounded on non-representational justifiers like headaches, then he can make sense of those justifiers being sufficiently available to you.

There Is Immediate Justification 215 Our next argument for the Premise Principle will question whether the foundationalist can make sense of the grounding relation, without appeal to beliefs or other representational states.

7 Grounding and Being Guided by Norms

We introduced the notion of a ground to distinguish between cases where you believe P for good reasons, or on grounds that justify you in believing P, and cases where you believe P on bad grounds, ones that do not justify that belief. What does it take for your belief to be grounded on some fact or condition C that you are in? A natural thought is that your belief counts as so grounded iff it is formed (or sustained) in a way that is guided by the epistemic norm “When in C, believe P.” If that is right, then the best way to understand the grounding relation is by inquiring into what it takes to be guided by such a norm. I understand an epistemic norm to be a claim about how we should be, in epistemic matters. Some norms merely evaluate the quality of a static epistemic situation, for example, You should not (it is inappropriate to) have inconsistent beliefs. Others instruct us how to change or improve our situation, for example, If you believe that A is F, then you should believe that B is F too; or You should gather as much evidence as possible. Only some epistemic norms tell us what to believe or to refrain from believing. Those are the norms that we need to consider here. We can take them to be of the form: When you are in conditions C, you should believe (or refrain from believing) P. Putting it in the imperative: When you are in C, believe P. Norms like these will be correct just in case being in C does make it the case that you should believe P. In other words, just in case being in C is a justification maker for the proposition that P. For any norm, there will be a difference between acting in a way that merely happens to accord with the norm, on the one hand, and being guided by the norm, or complying with it, on the other. You act in accordance with a norm “When in C, do ϕ” just in case you always ϕ when in C. You need not be trying to follow that norm. You may have ϕed for reasons that have nothing to do with C. You may even regard being in C as decisive reason to refrain from ϕing – but just never have noticed that you were in C. In order to comply with the norm, on the other hand, the fact that you are in C does in some sense need to guide or be your reason for ϕing. We need to know what this relation amounts to, when ϕing consists in forming (or sustaining) a belief. One account of this will portray you as deliberately following the norm, in the way that one can deliberately follow a cooking recipe. I mean three things by this. First, on this account your belief will be voluntarily chosen. Forming it will be a genuine action of yours. Second, your belief will be chosen for a reason (a practical reason). In forming the belief that P, you will have been guided by a norm “When in C, believe P” only if the fact (or apparent fact) that you are in C is among your reasons for actively forming that belief. Third, this account says that to be acting for the reason that you are in C, you have to represent that reason to yourself. You have to be in a position to employ the proposition that you are in C as a premise in your practical reasoning. The upshot of these three assumptions is that your belief can be guided by the norm “When in C, believe P” only when you represent to yourself that you are in C, and can employ that proposition as a premise in reasoning. This may be thought to lend some support to the Premise Principle.22

216 James Pryor Some replies. First, even if this account of belief-formation were right, it is not clear that it would really support the Premise Principle. It seems only to support the claim that, for C to be a justifier that grounds your belief, you need to be in some state that assertively represents you as being in C. This doesn’t imply that C itself is a representational state. Still, this account will imply that representational states are present whenever your belief is properly grounded. And that may be of some use to a defender of the Premise Principle. He may urge that it is really the state that represents you as being in C, and not the condition C itself, that is doing the justifying.23 Second, this account of belief-formation seems to rely on too reflective and deliberate a picture of what it takes to act for a reason, or in compliance with a norm. Consider activities like playing a musical instrument, figuring out why your car or computer isn’t working, or making judgments of grammaticality. These are practical skills whose exercise seems to be governed by rules. But we don’t think subjects need to think about or deliberately apply any rules when they are performing those activities. They don’t even need to be aware of why they are acting in the precise ways they do. Many philosophers would regard their actions as guided by rules, for all that. So, prima facie, it seems possible to act in a way that is guided by rules without representing to yourself that you are in conditions C so now you should do so and so.24 Finally, I think the present account of belief-formation misrepresents how active we are with respect to our beliefs. Many justified beliefs aren’t formed in the deliberate way it describes because forming those beliefs isn’t an action of ours in the first place. We do exercise voluntary control over some aspects of our epistemic lives: what evidence we gather, what sources we consult, and so on. But when it comes to our here- and-now doxastic choices, these are usually involuntary and unreflective. Our beliefs usually just result from our other epistemic efforts. They just happen, in the way that sneezing or digesting happen. Ordinarily we make no intentional choices about what to believe. One can choose to believe something, and then seek ways to get oneself to believe it – just as we can seek ways to get ourselves to sneeze or digest. But that is not the way we usually form justified beliefs.25 Some philosophers argue for the strong thesis that it is impossible for justified beliefs ever to be formed by deliberate choice. Perhaps that is right. But here I need only the following, much weaker thesis: some beliefs are appropriately held, and so properly grounded, even though they aren’t formed by deliberate choice. So it can’t in general be required for grounding a belief that one have formed the belief in the deliberate way that the account we are considering describes. Here a proponent of the Premise Principle might say: “True, we don’t always ourselves deliberately choose our beliefs. But that is how an ideal reasoner would form beliefs. And it is a constraint on any condition that purports to justify our beliefs that it could ground the belief of an ideal reasoner who did choose his beliefs deliberately.” I am not sure that the fully reflective and deliberate reasoner envisaged here is a coherent ideal for us to aim at. Such a reasoner would never form or change beliefs, except by deliberately following an epistemic norm. But deliberately following a norm requires already having beliefs about (or at least, some representation of) whether its antecedent conditions are fulfilled. From where is the ideal reasoner supposed to get those beliefs? Since she is an ideal reasoner, she would have to have formed them by deliberately following norms, too. But then she would need further beliefs, about

There Is Immediate Justification 217 whether the antecedent conditions of those norms are fulfilled.… There is a real threat that this reasoner would never be able to get started. She would never be able to deliberately follow a norm for forming beliefs. So she would be doxastically paralyzed. Or, if she were able to get started, it would only be by virtue of believing an infinite hier- archy of propositions. Neither option gives us a very promising model to aspire to.26 If your belief’s being grounded on condition C isn’t a matter of your deliberately following a norm when you form that belief, then what is it a matter of? This is a dif- ficult question. I can’t guarantee that when we work out the details, they won’t turn out to be inhospitable to views that violate the Premise Principle. But I think we can be assured that there is no argument here for the Premise Principle “from on high,” no argument that will settle the question before we work out those details.

Acknowledgments

I would like to thank Tony Brueckner, Rich Feldman, Mark Greenberg, Ram Neta, Christian Piller, Mark Schroeder, Matthias Steup, and audiences at NYU, UNC-Chapel Hill, Princeton, Kansas, Oxford, and York, for conversations and feedback that helped me write this essay.

Notes

1 Though it doesn’t include everything. I think it is possible for subjects to believe P inappropriately, and so without justification, though be non-culpable for doing so. (Perhaps the epistemic faults that led to the belief are too subtle and well entrenched for those subjects to recognize.) So there is one kind of positive epistemic status, being epistemically blameless in believing P, that does not entail that one has justification to believe P. See Pryor (2001, section 4) for discussion and references. 2 The relation of believing something for such-and-such a reason is sometimes called “the basing relation.” However, I think that terminology encourages too voluntaristic and reflective a picture of the phenomenon; I prefer to call it “the grounding relation” instead. We will talk more about this relation in section 7. For further discussion, see Korcz (1997) and Audi (1993, chapters 3, 7, and 8). 3 For more about immediate justification and what it does and does not require, see Alston (1989, chapter 3), Audi (1993, chapter 3), and Pryor (2000). 4 We have to be careful here: I expect that what makes you justified in believing one proposition may often make you justified in believing several. Suppose being in state S makes

you justified in believing P1, . . . , Pn. As I understand talk of “making it true” that you have

justification, being in S can make it true that you have justification to believe P2, . . . , Pn (let S* be the state of having this justification), and also make it true that you have justification

to believe P1, without its following that being in S* is part of what makes you justified in

believing P1. Hence, your justification to believe P1 can be immediate, even if what makes

you justified in believing P1 also makes you justified in believing other propositions, too. 5 Consider: in order to have the concept of a unicorn I may need to believe (i) that unicorns have hooves, and (ii) that unicorns have horns. Now suppose I acquire evidence that a virus has killed all hoofed creatures. Since I believe unicorns to be hoofed creatures, I form the belief (iii) that no unicorns currently exist. It is clear that (ii) plays no role in justifying this belief. This shows that there can be propositions you need to believe in order to have certain

218 James Pryor concepts (you need to believe (ii) in order to have the concept of a unicorn), without those propositions mediating your justification for every belief involving the concepts. Now, (iii) is not an immediately justified belief. But it serves to make my point. We can see the same phenomenon with beliefs that are good candidates to be immediately justified, like (iv) If any unicorn exists, it is identical with itself. (ii) plays no more role in justifying that belief than it plays in justifying (iii). 6 On some presentations of the Regress Argument, the regress stops with your last justified belief (or, as I say, with the last proposition you have justification to believe). On others, the regress stops with what justifies your last justified belief: the states or conditions that make you justified in believing it. These are just different ways of talking; there is no real philosophical disagreement here. 7 Many foundationalists shun option (iii), but structurally it also qualifies as a foundationalist option. For example, a foundationalist might claim that merely having a belief that P – even without justification – makes you justified in believing that you do have that belief. 8 I owe this example to Tim Maudlin. 9 Briefly, and I as I understand it, that theory invoked a mode of awareness that (i) had a success-grammar, (ii) was a cognitive relation to things or events, rather than to propositions, and (iii) wasn’t mediated by your awareness of anything else. This mode of awareness went under various names: direct apprehension, acquaintance, etc. It was usually claimed that what we are so aware of are sense data. However, as I understand the Given Theory, it neither entails, nor is it entailed by, belief in sense data. The Given Theory is undergoing something of a revival these days: see Fales (1996) and DePaul (2001). 10 Davidson and BonJour (when he was a coherentist) are pure coherentists; Cohen (2002) is a recent example of impure coherentism. 11 The argument dates back to a debate between Schlick and Hempel. Schlick said we can sometimes “compare propositions to facts,” and thereby acquire justification for believing those propositions. Hempel argued that the only way to acquire justification for believing a proposition is to “compare” it to things that stand in logical relations to it, namely other propositions . See Schlick (1932/3), Hempel (1934/5a), Schlick (1934/5), and Hempel (1934/5b). Wittgenstein voices a view akin to Hempel’s in Philosophical Investigations, section 486. The Master Argument has been given many times since (for example, Sellars, 1963, sections 3–7; Williams, 1977, chapter 2; BonJour, 1978, section 4, 1985, chapter 4; Davidson, 1986). (In more recent writings, BonJour rejects the Master Argument. See note 18, below.) 12 Some might want to count instrumental desires as epistemically justified when the means–end beliefs that motivate them are justified. I wouldn’t. But in any case, we can set instrumental desires aside and just consider cases where you desire that P for its own sake. Those desires have propositional content, and they don’t need to be epistemically justified. 13 One finds this view in Sellars (1963), Hintikka (1969), Dretske (1981, chapter 4 and 6), Evans (1982, chapters 5–7), Peacocke (1983, chapter 1), Searle (1983, chapter 2), Burge (1986), Lewis (1986), and in many places since. It is disputed whether experiences also have additional introspectible properties, beyond their propositional content; but there is broad agreement these days among philosophers of mind that they at least have propositional content. 14 John Broome suggested that we might also count intentions as assertively representing propositions; propositions about how we will act in the future. If so, then the Master Argument would give us no reason to exclude intentions from the ranks of justifiers, either. 15 They may be able to justify those beliefs indirectly, if their content is such that you can infer from its truth that you are likely to have certain beliefs or experiences. But the coherentist should have no objection to that. 16 What McDowell (1994) calls “The Myth of the Given” is the view that things without conceptual content can play a justification-making role. Curiously, McDowell (1994, p. 144)

There Is Immediate Justification 219 seems to allow that the fact that you are having a certain sensation or impression might, together with justified beliefs about how such sensations are reliably caused, justify you in beliefs about the external world. McDowell thinks this kind of epistemic role is too “indirect” to be fully satisfactory; but he does seem prepared to count it as a justifying relation. This sits ill with his otherwise thoroughgoing commitment to the Premise Principle. 17 Two caveats about Peacocke. First, he would put his point like this: experiences can stand in the kinds of rational relations cited in the Premise Principle even if their content is non-conceptual. He is reluctant to call those relations “inferential relations” except when both relata are conceptual. Second, Peacocke does not himself accept the Premise Prin- ciple. He thinks that non-representational states like pains can also justify beliefs (see pp. 254–255). 18 Although many foundationalists’ theories commit them to rejecting the Content Require- ment and the Premise Principle, it is rare to find much explicit and sustained discussion of these principles. BonJour is one author who does discuss them. In BonJour (1978, 1985, chapter 4), he endorsed the Master Argument; but he has changed his mind and now rejects the Premise Principle. He thinks there can be “descriptive relations” between a belief and a non-propositional state that make the belief justified when one is in the non-propositional state. See BonJour (2001, pp. 29ff.). Other authors who discuss the Premise Principle are: Millar (1991, chapter 4), Reynolds (1991), and Fales (1996, chapters 5–6). Fales’s terminology can mislead, but in essence his view is this: experiences do not themselves have propositional contents, and so can’t stand in the kind of “inferential relations” required by the Premise Principle. However, their phenomenal qualities do have a proposition-like structure, and we have a way to non-propositionally “apprehend” this structure. Fales thinks that is all that is needed to justify our perceptual beliefs. Millar does allow that experiences have propositional contents, but he thinks experiences are individuated by their phenomenal types rather than by their contents. He also thinks it is these phenomenal types that are epistemologically important. He says there are “quasi-inferential” links between phenomenal types and beliefs that experiences of those types make appropriate. These links, not the experiences’ content, explain why experiences justify the beliefs they do. 19 I am indebted to Mark Johnston for discussions of this possibility. 20 The noun “justification” has both a count use and a mass use. The count use (“He has a justification for that belief”) is most naturally read as referring to arguments or justification showers. The mass use (“He has some justification to believe that”; “She has more justification than he does”; etc.) is more naturally read as referring to the presence of justification makers. 21 I am indebted to Mark Schroeder for discussions of this objection. See also Unger (1975, chapter 5). BonJour (1985, chapters 2 and 3) and Brewer (1999, pp. 19, 49, and esp. 163ff.) claim that nothing can be a reason for you unless you are in a position to recognize it as a reason. McDowell (1994, pp. 52–53) insists that anything that is going to count as a reason- giving relation “must be able to come under the self-scrutiny of active thinking.” I think this means that your reasons must be available for you to think about and critically assess. 22 See, for example, Brewer (1999, pp. 165–169). Pollock considers, and criticizes, a similar argument on behalf of what he calls “the doxastic assumption” that only beliefs can be justification makers (Pollock, 2001, p. 41; see also Pollock and Cruz, 1999). 23 Brewer (1999, pp. 168–169) takes this line. 24 Other philosophers have also argued against overly deliberate and reflective accounts of being guided by an epistemic norm. See Pollock and Cruz (1999, pp. 124–130 and 136–137), Millar (1991, esp. p. 121), and Reynolds (1991). 25 For some recent discussion of the question whether belief is voluntary, see Alston (1989, chapter 5), Steup (2000), Feldman (2001), Audi (2001), and Ginet (2001). 26 See also Van Cleve (1979), Pollock and Cruz (1999, p. 125), and Pollock (2001, pp. 44–45).

220 James Pryor References

Alston, W. (1989) Epistemic Justification. Ithaca, NY: Cornell University Press. Audi, R. (1993) The Structure of Justification. Cambridge: Cambridge University Press. Audi, R. (2001) Doxastic voluntarism and the ethics of belief. In Knowledge, Truth and Duty: Essays on Epistemic Justification, Responsibility, and Virtue, ed. M. Steup. Oxford: Oxford University Press. BonJour, L. (1978) Can empirical knowledge have a foundation? American Philosophical Quarterly 15: 1–13. BonJour, L. (1985) The Structure of Empirical Knowledge. Cambridge, MA: Harvard University Press. BonJour, L. (2001) Toward a defense of empirical foundationalism. In Resurrecting Old-fashioned Foundationalism, ed. M.R. DePaul. Lanham, MD: Rowman & Littlefield. Brewer, B. (1999) Perception and Reason. Oxford: Oxford University Press. Burge, T. (1986) Cartesian error and the objectivity of perception. In Subject, Thought and Context, ed. P. Pettit and J. McDowell. Oxford: Oxford University Press. Chisholm, R.M. (1989) Theory of Knowledge, 3rd edn. Englewood Cliffs, NJ: Prentice Hall. Cohen, S. (2002) Basic knowledge and the problem of easy knowledge. Philosophy and Phenomenological Research 65: 309–329. Davidson, D. (1986) A coherence theory of truth and knowledge. In Truth and Interpretation: Perspectives on the Philosophy of Donald Davidson, ed. E. LePore. Oxford: Blackwell. DePaul, M.R. (2001) Resurrecting Old-fashioned Foundationalism. Lanham, MD: Rowman & Littlefield. Dretske, F. (1981) Knowledge and the Flow of Information. Cambridge, MA: MIT Press. Evans, G. (1982) The Varieties of Reference. Oxford: Oxford University Press. Fales, E. (1996) A Defense of the Given. Lanham, MD: Rowman & Littlefield. Feldman, R. (2001) Voluntary belief and epistemic evaluation. In Knowledge, Truth and Duty: Essays on Epistemic Justification, Responsibility, and Virtue, ed. M. Steup. Oxford: Oxford University Press. Firth, R. (1964) Coherence, certainty, and epistemic priority. Journal of Philosophy 61: 545–557. Ginet, C. (2001) Deciding to believe. In Knowledge, Truth and Duty: Essays on Epistemic Justification, Responsibility, and Virtue, ed. M. Steup. Oxford: Oxford University Press. Harman, G. (1986) Change in View. Cambridge, MA: MIT Press. Heck, R.G. Jr (2000) Non-conceptual content and the “space of reasons.” Philosophical Review 109: 483–523. Hempel, C.G. (1934/5a) On the logical positivists’ theory of truth. Analysis 2: 49–59. Hempel, C.G. (1934/5b) Some remarks on “facts” and propositions. Analysis 2: 93–96. Reprinted in Essential Readings in Logical Positivism (1981), ed. O. Hanfling. Oxford: Blackwell. Hintikka, J. (1969) On the logic of perception. In Models for Modalities. Dordrecht: Reidel. Korcz, K.A. (1997) Recent work on the Basing relation. American Philosophical Quarterly 34: 171–189. Lewis, D. (1986) Veridical hallucination and prosthetic vision. In Philosophical Papers, Volume 2. New York: Oxford University Press. McDowell, J. (1994) Mind and World. Cambridge, MA: Harvard University Press. McDowell, J. (1998) One strand in the private language argument. In Mind, Value, and Reality. Cambridge, MA: Harvard University Press. Martin, M. (1993) The rational role of experience. Proceedings of the Aristotelian Society 93: 71–88. Millar, A. (1991) Reasons and Experience. Oxford: Clarendon Press. Peacocke, C. (1983) Sense and Content. Oxford: Oxford University Press. Peacocke, C. (2001) Does perception have a non-conceptual content? Journal of Philosophy 98: 239–264. There Is Immediate Justification 221 Plantinga, A. (2001) Direct acquaintance? In Resurrecting Old-fashioned Foundationalism, ed. M.R. DePaul. Lanham, MD: Rowman & Littlefield. Pollock, J. (2001) Nondoxastic foundationalism. In Resurrecting Old-Fashioned Foundationalism, ed. M.R. DePaul. Lanham, MD: Rowman & Littlefield. Pollock, J. and Cruz, J. (1999) Contemporary Theories of Knowledge, 2nd edn. Lanham, MD: Rowman & Littlefield. Pryor, J. (2000) The skeptic and the dogmatist. Noûs 34: 517–549. Pryor, J. (2001) Highlights of recent epistemology. British Journal for the Philosophy of Science 52: 1–30. Pryor, J. (2004) What’s wrong with Moore’s argument? Philosophical Issues 14: 349–378. Reynolds, S.L. (1991) Knowing how to believe with justification. Philosophical Studies 64: 273–292. Schlick, M. (1932/3) Über das Fundament der Erkenntnis. Reprinted as On the foundation of knowledge. In Essential Readings in Logical Positivism (1981), ed. O. Hanfling. Oxford: Blackwell. Schlick, M. (1934/5) Facts and propositions. Analysis 2: 65–70. Reprinted in Essential Readings in Logical Positivism (1981), ed. O. Hanfling. Oxford: Blackwell. Searle, J. (1983) Intentionality. Cambridge: Cambridge University Press. Sellars, W. (1963) Empiricism and the philosophy of mind. In Science, Perception and Reality. London: Routledge & Kegan Paul. Steup, M. (2000) Doxastic voluntarism and epistemic deontology. Acta Analytica 15: 25–56. Steup, M. (2001) Foundationalism, sense-experiential content, and Sellars’s dilemma. http:// matthiassteup.files.wordpress.com/2013/01/sellarsian-dilemma1.pdf (last accessed March 23, 2013). Unger, P. (1975) Ignorance: A Case for Skepticism. Oxford: Oxford University Press. Van Cleve, J. (1979) Foundationalism, epistemic principles, and the Cartesian circle. Philosophical Review 88: 55–91. Williams, M. (1977) Groundless Belief. Oxford: Blackwell. Williamson, T. (2000) Knowledge and its Limits. Oxford: Oxford University Press.

There Is No Immediate Justification Juan Comesaña

Introduction

In his paper, Pryor argues for the existence of immediate justification and defends it against what he calls “the master argument for coherentism.” In this paper, I want to present and evaluate a different argument against one kind of immediate justification. My conclusion will be that, although the argument does not conclusively establish that there is no immediate justification in the cases to which it applies, it does show that the existence of that kind of immediate justification is incompatible with a set of widely accepted principles.

222 Juan Comesaña Where Does Immediate Justification Come From?

Pryor defines “immediate justification” as follows:

When your justification to believe P does not come from your justification to believe other propositions, I will call it “immediate.”

The phrase “coming from” sounds innocuously prosaic, but we will see that it must be functioning in the definition of immediate justification as a technical term. Right after presenting this characterization, Pryor makes five clarifications, two of which will be particularly important for our discussion. First, to put it in a common terminology, Pryor is talking about propositional rather than doxastic justification. Roughly speaking, propositional justification has to do with which propositions you are justified in believing, whether you believe them or not (and if you do believe them, whether you do so for good reasons or not); doxastic justification, on the other hand, has to do with whether your mental state consisting in believing a proposition is justified. We will be concerned with both propositional and doxastic justification. Second, Pryor believes that “justification is usually defeasible.” So, when he is talking about immediate justification, he is talking about prima facie justification, which can turn into all things considered justification if it is not defeated. Pryor thinks that the best defense of the existence of immediate justification comes not from some theoretical argument such as a version of the regress argument, but rather from consideration of cases. He thus gives examples of what he takes to be cases of immediate justification:

•• I feel tired, and I am thereby justified in believing that I feel tired. •• I have a headache, and I am thereby justified in believing that I have a headache. •• I raise my arm to scare a fly, and I am thereby justified in believing that I raise my arm to scare a fly. •• I imagine my grandmother sitting in her kitchen, and I am thereby justified in believing that I am imagining my grandmother sitting in her kitchen. •• I think about ways for a domino piece to cover two spaces on a chessboard, and I am thereby justified in believing that the only way to wholly cover two spaces on the board is to place the domino horizontally or vertically.

Pryor thinks that in each of these cases my justification for believing the relevant propositions does not come from my justification for believing any other proposition. A crucial question in what follows is: where does the justification come from? A good way to approach that question is to recall Pryor’s discussion of the master argument for coherentism. The version of the argument that Pryor settles on is the following:

Premise Principle: The only things that can justify a belief that P are other states that assertively represent propositions, and those propositions have to be ones that could be used as premises in an argument for P. They have to stand in some kind of inferential role to P: they have to imply it or inductively support it or something like that.

There Is No Immediate Justification 223 Only Beliefs: Only beliefs (or other states that are epistemically like beliefs) represent propositions assertively.

Therefore,

Only beliefs (or other states that are epistemically like beliefs) can be justifiers.

Pryor points out that there are reasons to doubt both of the premises in this argument. Against the Only Beliefs premise, many epistemologists hold that experiences are states that can represent propositions assertively without being epistemically like beliefs (that is to say, without needing – or indeed, allowing – to be justified in order to justify). When I have an experience as of hands in front of me, my experience assertively represents the proposition that there are hands in front of me, and that proposition could obviously be used as an argument for the conclusion that there are hands in front of me (although to call that an argument is to stretch the terminology to uncomfortable levels). Pryor also argues against the Premise Principle. I won’t be concerned here with the details of Pryor’s argument against the Premise Principle, but I will be concerned with exactly what aspect of the Premise Principle Pryor is rejecting. Let us now return to Pryor’s initial examples of immediately justified beliefs. Remember our question: if the justification in those examples doesn’t come from justification for believing anything else, where does it come from? One might be tempted to think that Pryor’s answer is the following: my justification for believing that I have a headache is my headache; my justification for believing that I feel tired is my feeling of tiredness, and so on. That is to say, in each of those cases, my justification comes from a state that doesn’t assertively represent any content. Note well: the idea here would be to say that my justification comes from the state itself, not from the fact that I am in that state, or from the event of my being in that state: it is my headache that justifies me in believing that I have a headache, not the fact that I have a headache or my having the headache. I said that this is what one might expect Pryor to say, but it is not what he actually says. What he actually says is the following:

Many foundationalists believe in justification-making facts that violate that constraint. For instance, many foundationalists want to allow facts about what sensations you are having, or facts about what mental activities you are engaging in, to count as justifiers.

According to Pryor here, it is not the states themselves that are the justifiers, but rather the fact that the subject is in such and such a state (or the event of the subject’s being in such and such a state – I won’t distinguish between these in what follows).1 According to this view, for instance, facts about the existence of certain experiences are immediate prima facie justifiers for belief in the propositions that are the contents of such experiences. Thus, the fact that I am having an experience with the content that there is a hand in front of me is an immediate prima facie justifier for the proposition that there is a hand in front of me. It is for this kind of view that I now want to raise a serious problem.

224 Juan Comesaña The Problem

The problem actually arises not only for that kind of view, but for a much larger class. To be susceptible to the problem, a view must satisfy four principles (eventually we will add a fifth), all of which are in fact widely accepted. The first principle in question I will call “Inductivism”:

Inductivism: A justifier J may provide justification for a subject S to believe that P even if J doesn’t entail that P.

Notice that, to satisfy Inductivism non-trivially, a view must hold that, at least sometimes, justifiers can be propositions (or facts or events, to the extent that we can make sense of the idea that facts and events can stand in logical relations to propositions).2 The view according to which facts about mental states can be justifiers satisfies Inductivism. To be sure, there will be cases where a justifier does entail the proposition that it justifies – for instance, that I have a headache may be where my justification for believing that I have a headache comes from. But there will be cases where the entailment fails, such as the one already mentioned where the fact that I have an experience with the content that there is a hand in front of me justifies me in believing that there is a hand in front of me. Of course, the proposition which is the content of the experience does entail the proposition that I am justified in believing (they are indeed the same proposition); but, according to the view we are now interested in, it is not the content of the experience, but the fact that I have it, which justifies me – and the fact that I have an experience does not entail that it is veridical. So, someone who believes in the existence of this kind of immediate justification will be committed to Inductivism. The converse, of course, doesn’t hold: one may be committed to Inductivism even if one doesn’t believe in the existence of immediate justification at all, let alone this specific kind of immediate justification. Second, a principle of Closure:

Closure: If S has justification for believing that P and competently deduces Q from P, then S has justification for believing that Q.

Closure principles have been at the center of the epistemological discussion for a while now.3 This is not the place for further substantive discussion about them, but I will make two points. First, I want to register my sympathy with defenders of Closure principles against their opponents. Nozick (1981) was right when he said (something to the effect) that we may quibble about the precise formulation of an exceptionless closure principle, but it is hard to deny that something like Closure must be right. (Nozick famously went on to deny closure, not on the basis of quibbling with the formulation of the principle, but rather on the basis of an analysis of knowledge.) Second, to generate the problem it is not necessary to hold that a principle like Closure holds with full generality. It is sufficient to hold that, in many particular cases, someone who is justified in believing a proposition P by a justifier which doesn’t entail P may competently perform a deduction from P to Q and thereby become justified in believing Q. To introduce the third principle, consider the following scenario. Suppose that I throw a fair die and hide the result. I then tell you that the die landed either on 1 or on

There Is No Immediate Justification 225 2. That gives you some evidence to believe that the die landed on 1, of course – you now have more justification than you did before for believing that the die landed on 1. And that the die landed on 1 entails that it didn’t land on 2. But think how absurd it would be if you reasoned this way: “I am now more justified than I was before in believing that the die landed on 1, and that the die landed on 1 entails that it didn’t land on 2. Therefore, I am now more justified than I was before in believing that the die didn’t land on 2.” Why would this reasoning be so absurd? One reason (maybe not the only one) is that the proposition that the die landed on 1 is a mere lemma. All the justification you have for believing that proposition comes from your justification for believing that the die landed either on 1 or on 2. Of course, that the die landed on 1 or on 2 does not give you more justification than you had before for believing that it didn’t land on 2 – on the contrary, it gives you more justification than you had before for believing that it did land on 2. Now, if you had independent justification for believing that the die landed on 1, then of course you might well be justified in believing that it didn’t land on 2. But given that all your justification for believing that the die landed on 1 comes from your justification for believing that it landed either on 1 or on 2, and given that this latter proposition does not justify you in believing that it didn’t land on 2, then you are not justified in believing that it didn’t land on 2. In slogan form, mere lemmas do not have any justifying power of their own. In principle form:

Mere Lemmas: If S’s justification for believing that P comes entirely from Q and Q does not justify S in believing R, then P does not justify S in believing R.4

Finally, suppose that P entails Q. Could Q then justify you in rejecting P? Hardly so. That Q is true is what would happen if P were true, so how can the truth of Q justify you in rejecting P? Our fourth principle enshrines this idea:

Entailment: If P entails Q, then Q cannot justify S in believing not-P.

Both Mere Lemmas and the Entailment principle are to be understood as restricted to contingent propositions – otherwise they would have counterexamples that are irrelevant for our purposes. As I said, many theories accept these four principles. Moreover, there seem to be very good reasons to accept these four principles. I have given brief defenses of each, and more will be provided below, but I think that they enjoy a great deal of initial plausibility. Unfortunately, they are incompatible with each other. There is a proposition P of a certain kind such that Inductivism and Closure entail you might be justified in believing P, whereas Entailment and Mere Lemmas entail that you can’t be justified in believing P. Let us start by assuming that the fact that you are having an experience with a certain content (let us say, that there is a hand in front of you, and let’s call this fact E) immediately justifies you in believing a certain proposition (let us say, that there is indeed a hand in front of you, “H” for short). As noted above, this case satisfies Inductivism non- trivially, for E does not entail H. Now, H entails the proposition that either H or not-E. Notice that this is equivalent to the proposition that it is not the case that E and not-H, and also equivalent to the material conditional that if E, then H. Let us suppose that you notice the entailment and competently deduce the proposition that either H or not-E. Closure now entails that you are justified in believing this proposition – although, as

226 Juan Comesaña noted above, nothing as strong as Closure is required to hold that you are indeed now justified in believing it. Now, a natural question to ask at this point is: what justifies you in believing that H or not-E? Or, to put it in Pryor’s terminology, where does your justification for believing H or not-E come from? There seem to be only two candidates in the offing: your justifier is either E or H itself (later I reexamine the assumption that these are the only two possible justifiers in this case). Take the latter option first. We are assuming that all your justification for believing H comes from E. The Mere Lemmas principle then entails that H can justify you in believing only things which E itself justifies you in believing. So, if you are justified at all in believing H or not-E, it must be because E itself justifies you. But notice that the negation of H or not-E is (equivalent to) E and not-H, which entails E. Therefore, according to the Entailment principle, E cannot justify you in believing H or not-E. So, according to the four principles under consideration, you are justified and it is not the case that you are justified in believing H or not-E. Therefore, the four principles are mutually inconsistent.5 Several variations of this problem have received attention in the recent literature. Something like it was proposed as a problem exclusive to reliabilism by Fumerton (1996) and Vogel (2000), who called it the “bootstrapping problem.” Fumerton and Vogel complained that reliabilism allows one to “bootstrap” perceptual knowledge into knowledge that perception is reliable. Something like it is also behind Wright’s worries with Moorean approaches to solving the skeptical problem (see, for instance, Wright, 2007). White (2006) also marshals a version of the problem against theories such as Pryor (2000), which hold that experience can give us immediate justification. Cohen (2002) argued that, properly understood, the problem afflicts not only reliabilism, but any theory which admits that we can have knowledge on the basis of a source without knowing that the source is reliable – he called it the “easy knowledge problem.” Independently of Cohen, Huemer (2001) argued that the problem is even more general, and it affects any theory which admits that there can be justification and knowledge on the basis of non-entailing evidence – he called it the “problem of defeasible justification.” Pryor (forthcoming) argues that White’s problem for dogmatism actually applies to any “credulist” theory – any theory according to which there can be what Pryor calls “non-quotidian” undermining. Recently, Sharon and Spectre (forthcoming) present a similar argument, but aimed against Closure.6 I cannot here present a detailed comparison between the problems presented by these authors and my own. Suffice it to say that I agree with Cohen, Huemer, and Pryor that there is a problem in the vicinity which affects a great variety of epistemological theories. Indeed, I think it is a virtue of my presentation of the problem that it reveals just how widely it applies. I disagree with Wright and Sharon and Spectre that the solution to the problems lies in denying some version of Closure. I also disagree with White, Cohen, and again Wright that the solution lies in claiming that we have a priori knowledge of the proposition in question. I turn now to precisely this issue.

The Neo-Rationalist Gambit and Immediate Justification

A crucial step in the argument that the four principles are mutually inconsistent is the claim that, in the cases under consideration, your justification for believing H or not-E can only come from two sources: either from E, or from H. Some philosophers will

There Is No Immediate Justification 227 deny this. They will say that you are a priori justified in believing that proposition – that is to say, you are justified but not on the basis of any empirical justifier such as H or E. Cohen (2010), for instance, holds a view of this sort and presents it as a solution to the easy knowledge problem. According to Cohen, given that E justifies H, a subject can assume E, defeasibly derive H, and then close the assumption by concluding that if E then H. The fact that this defeasible suppositional reasoning is easily available to any subject is what gives them a priori justification for believing that H or not-E, and so Cohen will block the argument presented in the previous section precisely at the point where we said that your justifiers for that proposition can only be E or H. Wedgwood (2012) holds a similar position. Wright (2004) and White (2006) also think that you have a priori justification for believing such propositions, but do not hold that what gives you this a priori justification is the kind of defeasible suppositional reasoning that Cohen advocates. Is this neo-rationalist position (neo-rationalist because it posits a priori justification for contingent propositions) a way of denying the existence of immediate justification? Not without additional assumptions. One can hold on to the claim that it is E alone which provides me with justification for believing H and also claim that I have a priori justification for believing H or not-E. One way of implementing this idea is by saying that it is E by itself which justifies me in believing H, whereas it is this very fact (that E justifies me in believing H), a fact which obtains independently of whether E itself obtains, that justifies me in believing H or not-E. But this position is not very stable. Remember that, according to Pryor, your justification for believing a proposition is immediate if and only if it doesn’t “come from” your justification for believing any other proposition. According to the neo-rationalist position, whenever some fact F gives you justification for believing a proposition P, you already have justification for believing the material conditional if F then P. If so, whenever you are justified in believing P on the basis of F you will have available to you a different justification – one which depends not only on F but also on the conditional if F then P. The justification in question will in addition be logically stronger – it will entail that P. This doesn’t mean, of course, that it is epistemically stronger, but it is nevertheless remarkable that, according to neo-rationalists, whenever there is immediate justification for a proposition there is also non-immediate justification for it. Of course, the friend of immediate justification is free to hold that even though you have justification for believing this conditional, your justification for believing the consequent “comes from” just the antecedent. This response, however, only highlights the obscurity in the phrase “comes from.” If we were talking about doxastic justification, there would be a natural explanation for the phrase: your doxastic justification for believing a proposition (if indeed you are doxastically justified) comes from whatever you based your belief on. Presumably, there is a fact of the matter about what your mental state of believing a proposition is based on, and we can appeal to this fact in explaining the phrase “comes from.” However, we are talking here about propositional justification, and no natural explanation of the phrase is available in this context. What determines whether your justification for believing P comes from (in part) the conditional or not? Perhaps one could try to argue for the claim that it doesn’t by saying that you would still be justified in believing P even if you were not justified in believing in the conditional. But remember that, according to the neo-rationalist, justification for the conditional will be available whenever the antecedent justifies the

228 Juan Comesaña consequent, and so the counterfactual in question will have an impossible antecedent. Many people believe that this makes the counterfactual trivially true. Alternatively, one could try arguing that what explains why you are justified in believing P in the circumstances is primarily F, and the conditional is just icing on the cake. Some such explanation, in any case, is indeed needed if one is to reconcile the neo-rationalist position with the existence of immediate justification. Whatever the prospects for a reconciliation of neo-rationalism with the existence of immediate justification, I will next argue that neo-rationalism cannot solve the problem we are interested in.

Grounding

The neo-rationalist denies that E and H are the only possible justifiers for H or not-E in the cases with which we are concerned. The position allows us to say the following: when you concluded that H or not-E based on your competent deduction from H, you didn’t acquire any justification that you didn’t have before. You were already justified in believing that proposition, and so the appearance of pulling yourself by your own bootstraps is neutralized. You would be pulling yourself by your own bootstraps if you were to gain justification to believe H or not-E by following that procedure, but you do not gain anything. This way of formulating the neo-rationalist position allows us to see that it is indeed about propositional justification. For instance, you need not engage in the suppositional reasoning that, according to Cohen, gives you justification for believing H or not-E; and you need not avail yourself of the justification that according to Wright is yours “by default.” Nevertheless, that justification is there available to you, and does give you propositional justification even if you do not ground your belief in it. Consider now a subject who doesn’t avail himself of that a priori justification. For all we know, this subject doesn’t even reflect on the proposition that H or not-E before he acquires evidence E. Let us further suppose that he does indeed come to believe H, and it is indeed E which justifies him in so believing. Now he goes ahead and competently deduces H or not-E from H. Is the subject then justified in believing that proposition? The question, of course, is ambiguous between the following two: (i) does the subject have propositional justification for believing that H or not-E?; (ii) does the subject have doxastic justification for believing H or not-E? It is crucial, in considering the answers to those questions, to reflect on the relationship between propositional and doxastic justification. Suppose that E gives me propositional justification for believing H. Suppose now that I do in fact believe H on the basis of E, and that nothing out of the ordinary is going on. Then, I am doxastically justified in believing H. Conversely, if I am doxastically justified in believing that H on the basis of E, then E must give me propositional justification for believing P. True, the clause that nothing out of the ordinary is going on may well hide interesting issues, but they are not issues that will affect our argument. I therefore propose the following principle:

Grounding: If S believes that Q based on P, then S is doxastically justified in believing that Q if and only if P provides propositional justification for S to believe that Q.

There Is No Immediate Justification 229 We should also distinguish two different versions of Closure, one applying to propositional and the other to doxastic justification:

Propositional Closure: If S is doxastically justified in believing that P and competently deduces Q from P, then S is doxastically justified in believing that Q on the basis of P.

Doxastic Closure: If S is doxastically justified in believing that P and competently deduces Q from P, then S has propositional justification for believing that Q.

Notice that, as I am understanding the Closure principles, the justification in question in the antecedent can only refer to doxastic justification – one may well notice that Q follows from P, but one can hardly deduce Q from P unless one already believes P.7 Principles similar to Doxastic Closure sometimes go under the name of “transmission” principles – but I do not want to commit myself on how closely related the different transmission principles and my Doxastic Closure are to each other. Let us now revisit our sample case. Given Inductivism and Grounding, you are doxastically justified in believing H. Now, according to Doxastic Closure, you are then doxastically justified in believing H or not-E. You believe H or not-E on the basis of H, and so Grounding entails that H must provide you with propositional justification for believing H or not-E. But, according to Mere Lemmas, that can only happen if E provides you with propositional justification for believing H or not-E, and according to the Entailment principle it cannot. Hence, the problem reappears.8 To recap: the original problem arises from the incompatibility of Inductivism, Closure, Mere Lemmas, and Entailment, plus the assumption that only E or H can justify you in believing H or not-E. Neo-rationalism challenges this assumption, positing a priori propositional justification for believing that proposition. But close attention to whether doxastic justification is possible in those cases shows that the problem can be reinstated: Inductivism, Doxastic Closure, Mere Lemmas, Entailment, and Grounding are mutually incompatible, and this time neo-rationalism has nothing to say about the problem. The existence of a priori propositional justification for believing a proposition is irrelevant to whether a subject who doesn’t avail himself of that justification is doxastically justified in believing that proposition.

Options

Let me now consider the prospects for denying each of the principles as a way out of the problem. Some philosophers have proposed that we reject Closure. For instance, Dretske (1970) and Nozick (1981) proposed theories of knowledge according to which closure principles for knowledge fail. Under plausible assumptions about the connection between knowledge and justification (plausible although, notably, not shared by Nozick and Dretske themselves), that means that our Closure fails. Most philosophers have taken these failures of closure principles to be powerful arguments against those theories, however. Other philosophers (notably, Wright) have held that although Propositional Closure is true, (something like) Doxastic Closure is false. As we saw, Wright concedes that if you are justified in believing H then you are also justified in believing H or not-E, but thinks that the justification for believing this latter proposition

230 Juan Comesaña does not come from your justification for believing H – it is rather available to you a priori. As far as I can tell, however, this combination of accepting Closure but denying transmission simply ignores the complications that arise when dealing with doxastic justification. According to Wright, you may well be doxastically justified in believing a proposition P, competently deduce Q from P, and yet fail to be doxastically justified in believing Q (let us suppose that you don’t avail yourself of the propositional justification for believing the Q in question available to you a priori, relying instead only on your impeccable deduction from P). Finally, whatever one thinks about Doxastic Closure, we must remember that, to generate the problem, we need not rely on any general Closure principle, but rather it is sufficient to hold that in particular cases a competent deduction from H to H or not-E can give you doxastic justification for believing this latter proposition. Klein (1995) has in effect denied Mere Lemmas. To be completely fair to Klein, he has defended the conditional claim that if we are to hold on to plausible closure principles, then Mere Lemmas has to go (although I think that Klein is clear that he does want to hold on to those closure principles, and so the stronger claim is also true). This also strikes me as a non-starter. To deny Mere Lemmas is to attribute magic powers to mere lemmas; it is to accept that epistemic justification can be created ex nihilo; more importantly, it is just to get the cases wrong. I anticipate that many philosophers friendly to the existence of immediate justification will want to deny the Entailment Principle. Indeed, Pryor (forthcoming) himself has taken this route. To me, the principle is undeniable. Pryor does two things. First, he points out the enormous class of cases for which the principle makes trouble. On this, I agree wholeheartedly. Second, he argues against one particular way of supporting the principle. I have something to say about this. Entailment can be supported by appeal to some probabilistic considerations. Suppose that you hold that E can justify P only if there is some probability function such that the conditional probability of P given E (in symbols, Pr(P | E)) is higher than the unconditional probability of P (Pr(P)). One may think this because one buys into a complete Bayesian package, according to which rational credences must be probabilities and must evolve according to the conditionalization rule.9 But one need not buy into the full Bayesian package to believe that there must be some probability function (not necessarily one which models the subject’s credences) according to which Pr(P | E) > Pr(P) (even if one doesn’t think that rational update of credences operates only by conditionalization) if one is to be justified in believing P on the basis of E. Now, it is a theorem of the probability calculus that if P entails E then Pr(not-P | E) < Pr(not-P) (provided that both P and E have non-extreme probabilities to begin with). Therefore, the Entailment principle follows from the probabilistic relevance constraint on justification. Attacking this probabilistic relevance, or perhaps some elements of its implementation, thus removes one important source of justification for the Entailment principle. But one need not buy into a probabilistic account of justification to find the Entailment principle very plausible. Indeed, when I presented the Entailment principle I made a brief argument in its defense which didn’t appeal to any probabilistic considerations. To repeat that argument in evidential terms, the idea is that if a certain proposition entails some evidence, then you cannot use that very evidence to reject the proposition. As far as that piece of evidence is concerned, everything is as it should be

There Is No Immediate Justification 231 when the proposition is true. Therefore, misgivings about probabilistic accounts of justification are not enough to dismiss the Entailment principle, which enjoys independent support.10 One could also deny the Grounding principle, which I used to reinstate the problem even for neo-rationalist positions. As it stands, the Grounding principle is certainly false. For instance, it is perfectly possible for me to have propositional justification for believing that I have no higher-order beliefs (beliefs about beliefs), but it is hard to see how I could ever be doxastically justified (in the way required by the Grounding principle) in so believing. Perhaps we should also take into account the manner in which I ground my belief in whatever provides me with propositional justification – for instance, if I believe that Q on the basis of Q and if P then Q but do it by applying the “rule” that from propositions of the form A and if B then C one can infer that C, then I don’t have doxastic justification.11 But the solutions to these problems, though perhaps not easy to state, shouldn’t lead us to think that there is nothing to the Grounding principle. The Grounding principle is true, details aside. The details are important, of course, but not likely to interfere with the use to which we are putting the principle here. Finally, we should take a careful look at Inductivism. Many philosophers would think it mad to give it up, for giving it up seems to invite skepticism and uphold the indefeasibility of justification. Although this is not the place to do it, I believe that a case can be made that abandoning Inductivism need not lead to either skepticism or indefeasibility. In a nutshell, we can avoid skepticism by holding that evidence that we usually do have does entail the propositions that we are justified in believing – for instance, we can hold that it is part of my evidence that I see that there is a hand in front of me. This position need not lead one to hold a “dijunctivist” view according to which our evidence differs from “good” to “bad” cases – roughly speaking, ordinary vs. skeptical scenarios – provided that one is comfortable with the possibility of false evidence, a possibility which I have argued for on independent grounds.12 When it comes to defeasibility, we should be careful not to confuse it with non-entailment. To para- phrase Pryor (2000), defeasible justification is justification which can be lost given improvements in one’s epistemic position.13 It is of course possible for one’s epistemic position to improve and, as a result, for one to lose justification to believe something entailed by one’s previous (and perhaps even present) evidence. Suppose, for instance, that you have a proof of some complicated theorem. You then show it to a very competent mathematician and (mistakenly) he assures you that there is a fallacy in the proof. You then cease to be justified in believing the theorem, even though you did have (and perhaps still have) evidence which entails it. Therefore, giving up Inductivism is irrelevant to holding on to the defeasibility of justification. The believer in the kind of immediate justification that I have been examining here is committed to Inductivism. But giving up Inductivism need not mean giving up on immediate justification. To be sure, some ways of giving up Inductivism are indeed ways of giving up on immediate justification. For instance, to put it in terms of our example, if one thinks (as some neo-rationalists may do) that justification for believing H or not-E is always involved in one’s justification for believing H on the basis of E, then one is thereby denying that E gives us immediate justification to believe H. But one may also think that my justification for believing that there is a hand in front of me is that there is a hand in front of me (although I only have this justification because

232 Juan Comesaña I have an experience with that content). This would count as immediate justification according to Pryor’s definition. The connection between Inductivism and immediate justification is therefore not straightforward.

Conclusion

Although, as just said, a commitment to immediate justification does not entail a commitment to Inductivism, many contemporary epistemologists are Inductivists because they hold that there is immediate justification which satisfies Inductivism. If, in addition, they also hold, as I have argued they should, that Closure (or better, particular applications of it), No Lemmas, and Entailment are true, then they are in trouble. There are at least two ways out of the trouble. They might give up on the existence of immediate justification altogether. Neo-rationalism is sometimes seen as a solution of this kind to the problem. I have raised doubts, however, both about neo- rationalism as a solution to the problem as well as about its supposed abandonment of immediate justification. The other way out of the problem consists not in giving up immediate justification, but in giving up inductivist immediate justification. This is the kind of solution that I think merits more attention than philosophers have given it so far.

Acknowledgments

Many thanks to Stewart Cohen, Jim Pryor, Carolina Sartorio, and Matthias Steup for comments on drafts of this paper. Thanks also to audiences at a Conference on Epistemology organized by the Center for the Study of Mind in Nature in Lofoten, Norway, in May of 2012 and a talk at the Sociedad Argentina de Análisis Filosófico, Buenos Aires, June 2012.

Notes

1 For relevant discussion, see Turri (2009). 2 What about the view that it is the states themselves which are the justifiers? Does that view satisfy Inductivism? On the face of it, it doesn’t, because states cannot enter into logical relations with propositions. Some philosophers may wish to hold that they do, in which case we would have to look into the details of their theories to see whether they satisfy Inductivism or not. Some of the issues here resemble those raised by “Sellars’s dilemma” – see Sellars (1956). 3 See the essays by Dretske and Hawthorne in Chapter 2 of this book. 4 Weisberg (2010) proposes a similar principle, which he calls “No-Feedback.” 5 Inductivism is a possibility principle: it says that it is possible for there to be a case of inductive justification. Together with Closure, it entails that it is possible for a subject to be justified in believing H or not-E when E is her evidence for H. But (subject to the pro- visos discussed in the next section) Entailment and No Lemmas entail that there is no such possible case. 6 I reply to Sharon and Spectre in Comesaña (forthcoming).

There Is No Immediate Justification 233 7 The reader may take the comment in the text as an indication of how I intend to use “deduce.” 8 For a similar argument, see Silins (2005). 9 The conditionalization rule has it that if E is all your evidence at a time t, then your confidence in any proposition P at t must be equal to the conditional probability of P on E. 10 For further arguments for Entailment, see my reply to Pryor in the essay below. 11 For relevant discussion, see Turri (2010). 12 See Comesaña and McGrath (forthcoming). 13 See Pryor (2000, p. 517). In that paper, however, Pryor sometimes slips and talks as if there were no difference between defeasible justification and non-entailing justification. For instance, on the very next page he talks of “defeasible justification, justification that does not guarantee that our beliefs are correct” (Pryor’s emphasis), and on the same page he talks of “defeasible, [i.e.] ampliative considerations.” Even more explicitly, footnote 11 on p. 543 includes the following: “The fallibilist’s central thesis is precisely that it’s possible to know things on the basis of defeasible evidence, evidence the possession of which is meta- physically compatible with your being wrong.”

References

Cohen, S. (2002) Basic knowledge and the problem of easy knowledge. Philosophy and Phenomenological Research 65 (2): 309–329. Cohen, S. (2010) Bootsrapping, defeasible reasoning and a priori justification. Philosophical Perspectives 24 (1): 141–159. Comesaña, J. (forthcoming) On Sharon and Spectre’s case against closure. Philosophical Studies. Comesaña, J. and McGrath, M. (forthcoming) Having false reasons. In Epistemic Norms, ed. C. Littlejohn and J. Turri. Oxford: Oxford University Press. Dretske, F. (1970) Epistemic operators. Journal of Philosophy 67; 1007–1023. Fumerton, R. (1996) Metaepistemology and Skepticism. Lanham, MD: Rowman & Littlefield. Huemer, M. (2001) The problem of defeasible justification. Erkenntnis 54: 375–397. Klein, P. (1995) Skepticism and closure: why the evil genius argument fails. Philosophical Topics 23 (1): 213–236. Nozick, R. (1981) Philosophical Explanations. Cambridge, MA: Harvard University Press. Pryor, J. (2000) The skeptic and the dogmatist. Noûs 34: 517–549. Pryor, J. (forthcoming) Problems for credulism. In Seemings and Justification: New Essays on Dogmatism and Phenomenal Conservatism, ed. C. Tucker. Oxford: Oxford University Press. Sellars, W. (1956) Empiricism and the philosophy of mind. Minnesota Studies in the Philosophy of Science 1: 253–329. Sharon, A. and Spectre, L. (forthcoming) Evidence and the openness of knowledge. Philosophical Studies. Silins, N. (2005) Transmission failure failure. Philosophical Studies 126: 71–102. Turri, J. (2009) The Ontology of Epistemic Reasons. Noûs 43 (3): 490–512. Turri, J. (2010) On the relationship between propositional and doxastic justification. Philosophy and Phenomenological Research 80 (2): 312–326. Vogel, J. (2000) Reliabilism leveled. The Journal of Philosophy 97 (11): 602–623. Wedgwood, R. (2012) A priori bootstrapping. In The A Priori in Philosophy. Oxford: Oxford University Press. Weisberg, J. (2010) Bootstrapping in general. Philosophy and Phenomenological Research 81 (3): 525–548.

234 Juan Comesaña Wright, C. (2004) Warrant for nothing (and foundations for free)? Aristotelian Society Supplementary Volume 78 (1): 167–212. Wright, C. (2007) The perils of dogmatism. In Themes from G.E. Moore: New Essays in Epistemology, ed. S. Nuccetelli and G. Seay. Oxford: Oxford University Press. White, R. (2006) Problems for Dogmatism. Philosophical Studies 131 (3): 525–557.

Reply to Comesaña James Pryor

(1) In my essay, I touched briefly on what I thought were the best reasons to think there is any immediate justification at all. But much of my discussion, there and elsewhere, is concerned not with that broad genus, but with two (overlapping) species of it: (i) immediate justification which is also fallible, defeasible, and especially underminable; and (ii) immediate justification where the source needn’t have a content that stands in an argumentatively nice relation to the proposition justified. In my essay, I spent the most time arguing, against the classical Davidsonian “Master Argument,” that the case for resisting justification of type (ii) is obscure. The “dogmatist” epistemology of perception that I’ve defended elsewhere doesn’t obviously require there to be immediate justification of type (ii). On many contemporary views of perceptual experience, and in my own initial thinking, experiences do have contents that stand in some kind of argumentatively nice relation to the propositions they justify. But I’ve come to think it is less important than I did before whether that is so. Moreover, I suspect that having a reasonable story about the ways in which perceptual justification can be undermined will independently require a dogmatist to overcome resistance to justification of type (ii). Comesaña’s essay also does not focus on the possibility of immediate justification quite generally, but rather on a particular species, roughly what I’ve here labeled type (i). He presents what deserves to be recognized as a second “Master Argument” in the direction of coherentism.1 The argument has several premises. Klein and Silins have entertained rejecting the Mere Lemmas principle, but most philosophers will agree with that premise.2 Closure is more controversial. As Comesaña points out, he doesn’t really need to rely on a Closure principle but only on the conditional “If you have justification for these premises, you also have justification for this consequence” being true in some particular cases. Some may argue that, even if we accept Closure, Comesaña’s examples still illustrate a failure of “warrant transmission” rather than the falsity of Inductivism. It is not clear to me whether that can be sustained in the dialectical setting Comesaña has presented, and, like him, I will not explore that strategy. I will focus on the conflict Comesaña highlights between Inductivism and his “Entailment principle.” As Comesaña says, this principle “enshrines” the idea that:

That Q is true is what would happen if P were true, so how can the truth of Q justify S in rejecting P?

Reply to Comesaña 235 I have discussed this motivating idea elsewhere.3 I agree the idea is intuitively compel- ling. If my friend Sandra tells me (Q) the FBI’s silence about captured aliens justifies her in thinking they’re hiding something, and hence in rejecting that (P) they don’t have any captured aliens, I want to protest: but if they didn’t have any, this silence is just what you’d expect! All the worse for Sandra’s argument if, as in Comesaña’s examples, the hypothesis being rejected entails the observations that allegedly tell against it. In the end, though, I think this idea is seductively false. Of course Sandra’s argument is a bad one; but I think the Entailment principle must be an overgeneralization about why. I don’t think it is an easy matter to persuade one of this, so all I can do here is try to motivate some second thoughts about the principle. I’ll also suggest that it’s not just the Inductivist that the principle threatens: some of the Inductivist’s opponents are in no position to embrace Comesana’s argument, either.

(2) Comesana proposes to ignore the non-contingent, but I can’t see how to assess the Entailment principle without looking at aspects of how entailment and justification interact that lie beyond where he directs our attention. Some of the doubts I’ll raise require you to attend to the difference between two kinds of case:

Case 1. You have some background justification B, which entails not-P, but doesn’t yet justify you in rejecting P. Case 2. You have the same background justification B, and acquire some extra evidence X, and B and X together do justify you in rejecting P.

If you agree that we sometimes lack justification for believing things entailed by our evidence, and that this lack can be remedied, then presumably you’ll think of the remedy as working like X does in Case 2.

(3) Now, Comesaña’s Entailment principle says: If P entails Q, then Q cannot justify S in believing not-P. I presume that whatever underwrites that principle would also underwrite the more general: If P entails Q (against the unchallenged background of B), then Q (together with B) cannot justify S in believing not-P. There may be complications here, especially if we seek a version of this principle applicable to degrees of confidence in P, but I will assume that in the kinds of cases Comesaña is considering, these principles stand or fall together.

(4) Take some conspiracy theory, like Sandra’s. Imagine the best possible evidence you can against her theory, which falls short of entailing its falsehood. If you are an Inductivist, you probably agree there is such evidence – you needn’t also subscribe to more controversial claims about fallible immediate justification. Now, if Sandra has any skill as a conspiracy theorist, you know what will happen. Her original theory will get supplemented with additional convolutions to explain why your evidence is just what we should expect. Imagine she takes a shortcut and just conjoins your evidence to her original theory. So now her new theory entails your evidence.

236 James Pryor Now, you only committed to that evidence telling against her original theory; perhaps you hesitate to say it tells against her new theory. So go gather evidence telling against the new theory: about why the combination of her old theory and the new conjunct are improbable and incredible. Let’s suppose this new evidence also falls short of entailing her theory is false. So she will conjoin that and present her third theory … If you’re an Inductivist, then I expect you’ll be inclined to think at some stage – if not already at the first – you will be justified in believing one of Sandra’s theories is false, even though her theory entails your evidence. (If not, I have some rare grue emeralds you may be interested in ….) Note that Sandra’s theories have none of the explanatory virtues of typical skeptical hypotheses. They only manage to entail all your evidence because they enumerate that evidence. Let me acknowledge one possible source of hesitation. If your grounds for disbelieving Sandra’s theory are entailed by her theory, you may hesitate to call those grounds “evidence” against her theory. But many Inductivists will still think it is possible to be justified in disbelieving Sandra’s baroque conspiracy. And you will have some grounds for so believing. Such grounds are all that Comesaña means by “evidence” or “justifier.” These Inductivists will agree that these grounds make it reasonable for you to reject a theory, whose truth would entail those grounds. In other words, they should think Comesaña’s Entailment principle here is false. Of course, Comesaña agrees this is what Inductivism commits one to (given his other premises). What I am trying to show is that you should feel no shame in that commitment. This is just a natural application of Inductivism, with no appeal to any of its more controversial extensions. There does seem to be some truth in the vicinity of Comesaña’s Entailment principle, as my reaction to Sandra’s opening argument evidenced. We haven’t yet identified what that is, and Comesaña can fairly hold that against us. But despite its initial allure, an Inductivist who thinks it’s possible to rea- sonably reject any of Sandra’s conspiracy theories, on some grounds, should feel secure in rejecting the Entailment principle. As I see it, the challenges here are how to reconcile our intuition that these grounds justify us with our reluctance to count them as “evidence”; and to identify what real truth it is that the Entailment principle wrongly generalizes.

(5) One response Comesaña may make is that Sandra’s theories must at some point become intrinsically incredible, given our background evidence. Perhaps the original theory was even inconsistent with that background – though if so, let’s suppose not in a way that’s immediately obvious. Comesaña may object that it is this intrinsic incred- ibleness, rather than any evidence, that justifies you in rejecting the conspiracy theories. Hence, these cases don’t undermine the Entailment principle. (Leaving that principle free to – in other cases – undermine Inductivism.) Even if Sandra’s theories were intrinsically incredible from the outset – and even if they were inconsistent with our background evidence – we are supposing this wasn’t initially obvious. Nor should it have been obvious. Instead, there was some first evidence X that, together with your background evidence, brought you to the point of being able justifiably reject one of her theories. This is kind of situation envisaged in (2), above. Now, if her theory at that point is P, perhaps you don’t think that your background evidence plus X are enough to justify you in rejecting P + X, though they

Reply to Comesaña 237 were enough to justify you in rejecting P. In the previous section, I said that even if an Inductivist hesitates at this stage, he is likely to think there’s some point, some evidence Y, such that your background evidence plus X plus Y are enough to justify you in rejecting P + X + Y. Perhaps in some sense, P will have already had to carry the seeds of its own rejection in it, for this to be so. (Or at least, the seeds of its own rejectability given B.) But those seeds needn’t be epistemically transparent to us. Later evidence like X and Y can be needed before we’re in a position to justifiably harvest them.

(6) Finally, let’s turn to the question of whom Comesaña’s argument threatens. In his essay, he discusses “neo-rationalists” who oppose a specific form of Inductivism, that takes our perceptual justification to be fallible and immediate. Comesaña discusses whether those Inductivist views are reconcilable with claims the neo-rationalists make. He also argues by appeal to his Grounding principle that the a priori justification the neo-rationalists posit doesn’t prevent the kind of difficulty he’s pressing against the Inductivists from arising. These issues are complex and I don’t agree with everything Comesaña says, but we can’t pursue it here. In any event, the kind of neo-rationalists he’s envisaging may well be Inductivists elsewhere – for example, about induction – so they should be as eager as their opponents to resist Comesaña’s Master Argument against Inductivism. Given the options we’re considering, that means resisting the Entailment principle. But what about a philosopher who is resolutely and thoroughly a rationalist – who rejects any form of Inductivism. Would such a philosopher be in a position to coher- ently embrace the argument? Not necessarily. There may be some non-Inductivist views, perhaps the one Comesaña expresses his own sympathy for, that the argument does not threaten. But I think other non-Inductivist views may want also to reject the Entailment principle. Seeing this requires keeping in mind the contrast described in (2), above. For given the kind of view we’re now considering, any counterexample to the Entailment principle must take this form: some evidence Q supports not-P in a non-Inductive way, and hence entails not-P (at least, against your background evidence B). At the same time, P (against B) entails Q. Hence, P must already be inconsistent with B. However, as we’ve already noted, that doesn’t necessarily mean B already justified you in rejecting P. It may be that only B plus Q do that. If we don’t forget the difference between entailing and justifying, I expect that some rationalist views will want to diagnose our epistemic position with respect to Sandra’s conspiracies in just this way. More generally, the issue is that even reason-based evidence can be undermined. One can acquire evidence that at least makes it likely that one’s reasons are unsupportive, or are not competently in hand. If our rationalists are going to be non-skeptics, they’ll probably want our use of reasoning faculties to make some justified headway against skeptical hypotheses that threaten to undermine them in those ways. Like the Inductivist, they may want to do so even when those hypotheses entail whatever it is that the rationalist deems is our evidence. And as we’ve observed, our having justification to reject the skeptical threats may sometimes require this, even in cases where the threats entail their own falsity.

238 James Pryor Notes

1 As Comesaña notes, Michael Huemer has also stated, but not endorsed, something like this argument. 2 See Silins (2005, section 3.2). 3 See principle “NFI” in my “When Warrant Transmits” (Pryor, 2012).

References

Pryor, J. (2012) When warrant transmits. In Wittgenstein, Epistemology and Mind: Themes from the Philosophy of Crispin Wright, ed. A. Coliva, pp. 269–303. Oxford: Oxford University Press. Silins, N. (2005) Transmission failure failure. Philosophical Studies 126: 71–102.

Reply to Pryor Juan Comesaña

The meat of Pryor’s reply is what he takes to be a counterexample to Entailment. My main objective in this reply is to show that Entailment survives a proper account of Pryor’s case. Before doing that, however, I briefly address some other points. Pryor thinks that Entailment is related to other principles. He draws our attention to his discussion of principle NFI in Pryor (2012), and also considers a version of Entailment not restricted to contingent propositions. Interesting as they are, I will set those other principles aside, for I do not see how they are related to my argument. Entailment is not obviously related to Pryor’s NFI, which states that if E is compatible with H’, which is an alternative to H, then E cannot justify believe in H. NFI is a direct rejection of Inductivism, Entailment is not. At the end of my essay, I urged on the reader the possibility of rejecting Inductivism as a solution to the puzzle I presented. But this, of course, doesn’t mean that Entailment itself entails a rejection of Inductivism – it does so only in conjunction with Closure and No Lemmas. Pryor shows also, toward the end of his reply, that if Entailment is not restricted to contingent propositions then puzzles arise regarding cases where our (justified) lack of logical omniscience shows up. Indeed, that is precisely why I restricted Entailment to contingent propositions. I do not think, therefore, that the kind of case with which Pryor ends represents an objection to the argument of my essay.1 Now, on to what Pryor takes to be a counterexample to Entailment. This is the case of Sandra the conspiracy theorist, who, when faced with evidence against her theory, simply incorporates that evidence as a conjunct of her new theory. Her revised conspiracy theory now entails the evidence that went against her original conspiracy theory. Therefore, according to Entailment, we cannot use that evidence to reject the revised conspiracy theory. But surely we are justified in rejecting even her revised conspiracy theory. Therefore, Pryor argues, we should reject Entailment.

Reply to Pryor 239 But there is a gap in that argument (Pryor is perfectly aware of this gap). We can all agree that we are justified in rejecting Sandra’s revised conspiracy theory. Entailment is consistent with this claim. What we cannot do, if Entailment is true, is reject Sandra’s revised conspiracy theory on the basis of evidence which the theory itself entails. Still, Pryor may now ask a fair question: if it is not on the basis of the evidence which the theory entails, on which basis do we reject Sandra’s revised conspiracy theory? In what follows, I first offer arguments for thinking that Entailment gives the correct verdict in Pryor’s case, and then answer that fair question.2 What were your opinions about Sandra’s different conspiracy theories before you encountered them and the evidence that allegedly tells against them? Now, perhaps you never thought about those conspiracy theories before Sandra told you about them, so the answer to my question is “Nothing.” But I do not really mean “before” in a chronological sense, but rather in a logical sense. I want you to think about your opinions about Sandra’s conspiracy theories bracketing the relevant evidence that you now have. To illustrate the point, consider the fact that you now have evidence that Pryor thinks that Entailment is false. Even if, before reading Pryor’s reply, you had never considered whether Pryor liked Entailment or not, you can now reflect on how credible it was that Pryor would reject Entailment bracketing the evidence that his reply gives you. I want you to do the same with Sandra’s conspiracy theories. Because it simplifies exposition, I will talk as if the chronological interpretation of “before” were correct, but nothing hangs on this. Sandra has two conspiracy theories.3 Let’s call her first conspiracy theory not-H, because it denies some widely accepted hypothesis H. The revised one is the conjunction of not-H with evidence that overwhelmingly tells in favor of H. Let’s summarize that evidence as saying that everything tells in favor of H (short of entailing H, something which I take for granted in what follows). So, Sandra’s original theory is not-H and her revised theory is not-H but everything tells in favor of H. You still don’t know whether everything tells in favor of H is true or not, but you start wondering about possible conspiracy theories. You first consider not-H, and rationally have a very low opinion of it (it is a conspiracy theory, after all).4 But what should your opinion of not-H but everything tells in favor of H be? I take it that it is obvious that it should also be very low (but I examine Pryor’s alternative take on this later on). After all, if everything tells in favor of H is true, then most likely H is true. Now, everything tells in favor of H doesn’t entail that H, so you may want to leave some room for the possibility that the evidence is massively misleading and not-H is true but everything tells in favor of H is also true – but that room can be very, very small. So, even before knowing that everything tells in favor of H is true, your opinion of not-H and everything tells in favor of H should be very low. Now consider what should happen when you do learn that everything tells in favor of H is true. According to Pryor, what happens is that your confidence in not-H but everything tells in favor of H goes down enough for you to reject it on the basis of everything tells in favor of H. However, several interrelated considerations suggest that, if anything, your confidence in not-H but everything tells in favor of H should go up – although maybe just a tiny bit up. Now, if those considerations are convincing, then of course you cannot rationally reject Sandra’s revised conspiracy theory on the basis of that evidence, and so the verdict of Entailment for that case is vindicated. I offer here three of those considerations. First, probabilistic construals of evidence have it that your confidence in a conjunction should go up when you learn that a

240 Juan Comesaña conjunct is true. Like Pryor, however, I don’t want to put too much stock on probabilistic construals of evidence. But there are non-probabilistic considerations that point in the same direction. Second, consider the fact that, when you learn that everything tells in favor of H is true, you have eliminated one way in which not-H but everything tells in favor of H could have been false. How could eliminating one way in which a proposition can be false not give you more confidence in that proposition? True, you have also eliminated one way in which H and everything tells in favor of H could have been false, and so your confidence in this proposition also goes up. Indeed, your confidence in this latter proposition should rise much more than your confidence in the previous one (that is why you end up justified in believing H, after all). But still, your confidence in a conjunction cannot but be raised (if only a tiny bit) when you find out that a conjunct is true. Finally, here is what I take to be a decisive argument for the claim that learning that everything tells in favor of H is true should make your confidence in Sandra’s revised conspiracy theory go up. Everyone should accept that it is not the case that everything tells in favor of H is good evidence for rejecting Sandra’s revised conspiracy theory. After all, it is not the case that everything tells in favor of H entails that not-H but everything tells in favor of H is false (and I am supposing here that you know this to be so). But Pryor wants to claim that everything tells in favor of H also constitutes good grounds for rejecting not-H but everything tells in favor of H. That is, Pryor is committed to saying that a proposition and its negation can both be evidence for the same hypothesis (in this case, that Sandra’s revised theory is false). This cannot be right. To illustrate the implausibility of saying that both E and not-E can provide evidence for the same proposition, suppose that you are wondering whether a hypothesis is true, and know that you are facing one of those situations where both E and not-E provide evidence for the hypothesis in question. It seems that, absurdly, you should increase your confidence in the hypothesis even before acquiring any evidence. Indeed, if both E and not-E are evidence for the hypothesis, then why isn’t your confidence in the hypothesis higher to begin with? You know that, no matter which of E or not-E you learn, you will rationally increase your confidence in H (and not for tricky reasons, such as because you will forget relevant information). So why not increase it right now and not go to the trouble of inquiring into E? Those considerations can be further generalized. Carolina Sartorio and I have argued, on independent grounds, for the following principle (Comesaña and Sartorio, forthcoming):

Epistemic Difference-Making (EDM): If E is evidence for H, then not-E is not evidence for H.

Notice that an obvious instance of EDM is that if not-E is evidence for H, then E is not evidence for H. EDM entails Entailment (and, therefore, obviously entails that Entailment gives the right verdict in Pryor’s case). For suppose that H entails E. In that case, of course, not-E is evidence for not-H. Given EDM, E cannot therefore be evidence for not-H as well. I assume that to reject a hypothesis on the basis of some evidence the evidence must count against the hypothesis. Therefore, if H entails E, E cannot be used to reject H – that is to say, Entailment is true.

Reply to Pryor 241 Having defended Entailment and the verdict that it gives in Pryor’s case, let me go back to the fair question: if it is not the information that everything tells in favor of H is true that allows us to reject Sandra’s revised conspiracy theory, what exactly does allow us to reject it? Well, it is simply the fact that it is so unlikely to be true. I said that the truth of everything tells in favor of H is some evidence for not-H but everything tells in favor of H, but that leaves it open that it can be extremely flimsy evidence. So, given that your previous confidence in Sandra’s revised conspiracy theory was extremely low, and given that it has increased only by a very tiny margin, your confidence in that theory is still extremely low – low enough to justify you in continuing to reject it. As Pryor puts it, it is the “intrinsic incredibility” of Sandra’s revised conspiracy theory that allows us to reject it. Pryor anticipates this reply, and complains as follows:

Even if Sandra’s theories were intrinsically incredible from the beginning – and even if they were inconsistent with our background evidence – I’m supposing that this wasn’t initially obvious to you. Nor should it have been obvious to you. Instead, there was some evidence X that, together with your background evidence, brought you to the point of being able to justifiably reject one of her theories.

I will ignore, as I said at the beginning, Pryor’s flirtations with cases where the hypothesis in question are inconsistent with what you know. So, what Pryor is claiming is that even if not-H but everything tells in favor of H is initially incredible, this shouldn’t have been obvious to you, but rather something that you come to realize with the help of everything tells in favor of H.5 I take it that this means that you start out rationally having some high-ish confidence in not-H but everything tells in favor of H. One immediate puzzle is then the following: what is Pryor admitting when he admits that not-H but everything tells in favor of H might nevertheless be initially incredible? Incredible to whom? More importantly: if you do rationally have some high-ish confidence in not-H but everything tells in favor of H, how can everything tells in favor of H justify you in believing H? But let us grant for the sake of argument that it is indeed the case that you rationally have some high-ish confidence in not-H but everything tells in favor of H (remember that everything tells in favor of H is just my name for the mass of evidence that you present in favor of H). How could this confidence be rationally shaken by acquiring the information that everything tells in favor of H is true? There are two possibilities here. One possibility is that when you learn that everything tells in favor of H is true you realize that your confidence in not-H but everything tells in favor of H should have been low all along, even before acquiring the evidence. If so, this is more akin to a conversion than to a rational accommodation of evidence. More importantly, it doesn’t go against Entailment. What you would be doing is to retroactively lower your previous confidence in not-H but everything tells in favor of H, and then raise it a tiny bit, leaving it low enough to justifiedly reject that hypothesis. The second possibility, which must be what Pryor has in mind, is that learning that everything tells in favor of H is true rationally lowers your confidence in not-H but everything tells in favor of H. This position, however, clashes head to head with the arguments for Entailment that I presented above. If Pryor’s verdict about his case were obvious, that would constitute a good argument against Entailment nevertheless

242 Juan Comesaña (we would then have to scramble to see what to say about the previous arguments for Entailment). Alas, I do not find Pryor’s verdict obvious. I conclude, therefore, that Pryor’s case does not after all constitute a counterexample to Entailment.

Notes

1 Pryor mentions Huemer (2001) as someone who states, but doesn’t endorse, the same argument that I present in my essay. I greatly admire Huemer’s paper, but there are differences between Huemer’s argument and my own, and further differences as to what to do with the argument. For instance, Huemer shows only an intermittent grasp of the importance of No Lemmas, flirting with the Kleinian position I criticize in my paper. Huemer also favors a kind of contrastivist rejection of Closure that I don’t agree with, and conflates defeasible with inductive justification. 2 In line with the proposal at the end of my paper to take a closer look at life without Induc- tivism, I would urge the reader to consider the possibility that we have evidence that entails that Sandra’s theories are false. In that case, Entailment no longer applies (for Sandra’s theories can entail our evidence only at the cost of being inconsistent) and we are free to reject Sandra’s theories on the basis of that evidence. In what follows, however, I ignore this line of thought and consider what a more traditional answer would look like. 3 Pryor has Sandra build ever more revisions in the face of new evidence. I don’t think that going beyond two theories matters. 4 In what follows I talk in terms of degreed doxastic notions, such as low or high opinions and levels of confidence. This makes exposition easier, but I believe that the points I’m making survive translation to a coarse-grained approach to the doxastic attitudes. 5 Again, Pryor seems to think that something hangs on further iterating the process, but I don’t see that it does.

References

Comesaña, J. and Sartorio, C. (forthcoming) Difference-making in epistemology. Noûs. Huemer, M. (2001) The problem of defeasible justification. Erkenntnis 54: 375–397. Pryor, J. (2012) When warrant transmits. In Wittgenstein, Epistemology and Mind: Themes from the Philosophy of Crispin Wright, ed. A. Coliva, pp. 269–303. Oxford: Oxford University Press.

Reply to Pryor 243